Page 1

July–September 2017

The Official Journal of the International Association for Human Resource Information Management


Artificial Intelligence The power to improve efficiency and decision-making for HR professionals: from recruiting, to learning and development, to traditional HR operations.

See Mid-Year Buyer’s Guide Page 32-33


Volume 8, Number 3 • July-September 2017

From the Editors 3 Eric Lesser, Roy Altman, and Jeff Higgins

featured articles in this issue… What to do when Machines do Everything


By Ben Pring, Cognizant Technology Solutions Technology is no longer the domain of the few, but the province of the many. Those who win in the next phase of the digital economy are not necessarily those who can create the new machines, but those who figure out what to do with them.

AI and Shared Services Practical Applications in Business


By Shel Holtz, Holtz Communication + Technology Human resources experts anticipate artificial intelligence playing a role in keeping new hires informed and engaged between the time an offer is made and the employee’s first day on the job. Artificial intelligence will also help with onboarding, answering questions, and delivering timely updates.

Trading Cancer for Data: Machine Learning for Cancer Diagnosis


By Lior Gazit, Memorial Sloan-Kettering Cancer Center The future is happening and machines are leading the way for us. These are opportunistic times and the means necessary to seize this opportunity do not discriminate. You don’t need to belong to a specific gender or race, and your physical strength is irrelevant. Everyone can read, learn, and adapt. The future will serve us better, keep us safer, and make us healthier.

Managing a Human/Robotic Workforce – It’s Closer than You Think! 16 By Roy Altman, Memorial Sloan Kettering Cancer Center Although technology advances at revolutionary rates, human nature progresses at glacial, evolutionary rates. Many of humanity’s problems won’t be solved by advanced technology. Technology is a tool, which can be used to make life better or to manipulate and oppress others. Managing advanced technology is a tremendous burden; it is, in effect, managing the future of humankind.

The Workforce of the Future: HR’s Role in Managing the Amoeba

IBM Study: More than Half of CHROs See Cognitive Computing as a Disruptive Force in the Next Three Years 24 By Janet Mertens, IBM Institute for Business Value Cognitive capabilities can further advance the evolution of HR by expanding human expertise and improving employee experience. This study explores how key functions of HR can benefit from cognitive solutions, and highlights companies that are already leveraging cognitive capabilities to strengthen the employee experience and improve HR service delivery.

Artificial Intelligence: Ethics, Liability, Ownership and HR 27 By John Sumser, HR Examiner The age of human-machine integration is in its infancy. It is inevitable. In the transition, it is important that we move forward carefully with a clear picture of the risks and ethical issues. This article discusses these issues in detail and is a worthy starting point if you are considering the utilization of intelligent machines in your HR/Operations processes.

Bots ‘R Us


By Freddye Silverman, Silver Bullet Solutions One of the most critical factors in selection and deployment of any HR system now is the employee experience. It is expected to match or beat the consumer experience we have at home, both in ease and timeliness. Artificial intelligence, chatbots, virtual reality, drones, and holographs – all of these will contribute to an incomparable and highly personalized user experience. Take that, Amazon!

Executive Interview 37 An IHRIM WSR interview with Catharina Lavers Mallet, COO and head of product development for Talla Ms. Mallet provides unique insights and lessons learned from an emerging startup that is rolling out chatbots within the HR function.

The Back Story


May the Bots be with you: RPA for HR


By Al Adamsen, Talent Strategy Institute The workforce of the future cannot be thought of independently from those other people, those other entities, and those other things (robots, AI, etc.) doing work on behalf of an organization. All of them affect the employee experience, the organization’s culture, its brand, and its overall efficiency, effectiveness, and success.

By Katherine Jones, Ph.D., Mercer With the current emphasis on artificial intelligence in HR today as only the beginning, the ability to increasingly automate repetitive, transactional tasks will free professionals to address more strategic activities.

Page 32-33

Mid-Year Buyer’s Guide Workforce Solutions Review (ISSN 2154-6975) is published quarterly for the International Association for Human Resource Information Management by Futura Publishing LLC, 12809 Shady Mountain Road, Leander, TX 78641. Subscription rates can be found at www. Please send address corrections to Workforce Solutions Review at the address above. • Workforce Solutions Review • July-September 2017


Volume 8, Number 3 • July-September 2017

Workforce Solutions Review is a publication of the International Association for Human Resource Information Management, whose mission is to be the leading professional association for know­ledge, education and solutions supporting human capital management. Opinions expressed herein are not necessarily those of the editors, the IHRIM board of directors or the membership.

SCOTT BOLMAN, HR Transformation & HR Excellence Practice, Deloitte,

© 2017 All rights reserved

ELENA M. ORDÓÑEZ DEL CAMPO, Senior VP Global Delivery Unit, SAP AG, Frankfurt, Germany elena.


YVETTE CAMERON, Global Vice President Strategy, SuccessFactors, Littleton, CO LEW CONNER, Executive Director, Higher Education User Group, Gilbert, AZ USA

Managing Editor

GARY DURBIN, Chief Technology Officer, SynchSource, Oakland, CA USA

BRUNO QUERENET, Head of HR Technology/HR Operations, Genentech,

Dr. CHARLES H. FAY, Professor, School of Management & Labor Relations, Rutgers University, Highland Park, NJ USA

Co-Managing Editor

DR. URSULA CHRISTINA FELLBERG, Owner & Managing Director, UCF-StrategieBeraterin, Munich, Germany

MICHAEL RUDNICK, Managing Partner, Prescient Digital Media,

Associate Editors ROY ALTMAN, HRIS Manager - HR Analytics & Application Architecture at Memorial Sloan-Kettering Cancer Center, New York, NY SHAWN FITZGERALD, Managing Director, Total Rewards and HR Technology, Blue Cross Blue Shield Association, Chicago, IL, USA DAVID GABRIEL, ED.D., Global Reach Leadership, Berkeley, CA USA, JEFF HIGGINS, CEO, Human Capital Management Institute, Marina Del Rey, CA USA ERIC LESSER, KPMG, MICHAEL H. MARTIN, Partner, Aon Hewitt Consulting, Organization & HR Effectiveness, New York, NY DARSHANA NARAYANAN, Head of Research, pymetrics, New York, NY USA

EDITORIAL ADVISORY BOARD CECILE ALPER-LEROUX, VP Product Strategy and Development, Ultimate Software, Weston, FL MARK BENNETT, Work Life and Collaborative Products Strategy Director, Oracle Corp., Redwood Shores, CA USA ERIK BERGGREN, VP of Research, IDC, San Mateo, CA USA

ALSEN HSEIN, President,Take5 People Limited, Shanghai, PRC CARL C. HOFFMANN, Director, Human Capital Management & Performance LLC, Chapel Hill, NC USA JIM HOLINCHECK, VP Customer Deployment Applications, Workday, Chicago, IL USA

DR. DANIEL SULLIVAN, Professor of International Business, University of Delaware, Newark, Delaware USA MARK SMITH, CEO, Chief Research Officer, and Founder of Ventana Research, San Ramon, CA USA DAVE ULRICH, Professor, University of Michigan, Ann Arbor, MI USA JULIE YOO, Founder and Chief Data Scientist, Pymetrics, DR. MARY YOUNG, Principal Researcher, Human Capital, The Conference Board, New York, NY USA

IHRIM BOARD OF DIRECTORS Officers and Executive Committee JIM PETTIT, HRIP, SHRM-SCP, Chair MICK COLLINS, Vice Chair GARY MORLOCK, HRIP, CFO, Finance Committee Chair JOYCE BROWN, Secretary SHAFIQ LOKHANDWALA, Executive Director

CATHERINE ANN HONEY, VP Strategic Partner Relations, Safeguard World International, Boston, MA USA

Board Members

DR. KATHERINE JONES, Partner and Director of Research, Mercer, San Mateo, CA USA

STUART RUDNER, Director of Meeting & Events

SYNCO JONKEREN, VP, HCM Applications Product Development & Management, EMEA, The Netherlands MICHAEL J. KAVANAGH, Professor Emeritus of Management, State University of Albany (SUNY), Albany, NY USA BOB KAUNERT, Principal, Deloitte, Philadelphia, PA USA DAVID LUDLOW, Global VP, HCM Solutions, SAP, Palo Alto, CA RHONDA P. MARCUCCI, CPA, Consultant for GruppoMarcucci, Chicago, IL USA LEXY MARTIN, Independent Consultant/Researcher, Meadow Vista, CA BRIAN RETZLAFF, VP of Information Technology, Voya Financial, Atlanta, GA USA LISA ROWAN, Program Director, HR, Learning & Talent Strategies, IDC, Framingham, MA USA

JOSH BERSIN, Principal and Founder, Bersin by Deloitte, Oakland, CA USA


LISA STERLING, Executive Vice President, Chief People Officer, Ceridian, Lincoln, NE USA,

July-September 2017 • Workforce Solutions Review •

CATHERINE HONEY, Director of Member Services & HRIM Foundation Director MARYANN MCILRAITH, Director of Communities DOUG SAMPSON, Director of Marketing & Communications SHARON THOMPSON, HRIM Foundation Director

PUBLISHING INFORMATION TOM FAULKNER, Publisher, Futura Publishing LLC, Austin, TX USA, PATTY HUBER, Advertising Manager, Austin, TX USA

from the editors Eric Lesser, Lead Editor Eric Lesser recently joined KPMG as an executive director, Human Resources, with responsibility for workforce and human resources strategy. Prior to joining KPMG, he was the research director and North American leader for the IBM Institute for Business Value (IBV) where he led a global team of more than 50 professionals responsible for driving IBM’s research and thought leadership across a range of industry and crossindustry topics. In addition to setting direction and providing oversight across the IBV research portfolio, his recent publications focused on the future of workforce and human capital, analytics, social business, and enterprise mobility. Lesser is a co-author of Calculating Success: How the New Workplace Analytics Will Revitalize Your Organization (Harvard Business Review Press, 2012). He has also edited (with Laurence Prusak) Creating Value with Knowledge: Insights from the IBM Institute for Business Value (2003). He also edited Knowledge and Social Capital and co-edited Knowledge and Communities (2000). In addition to more than 20 white papers, he has written numerous articles for publications such as the Sloan Management Review, The Academy of Management Executive, the International Human Resources Information Management Journal, and the Journal of Business Strategy. He has been quoted in numerous publications, including The Wall Street Journal, BusinessWeek, the Financial Times, USA Today and the Chicago Tribune, and has appeared on Fox Business News, BBC News, and CBC Newsworld. He speaks frequently on a variety of human capital topics. He can be reached at Roy Altman, Contributing Editor Roy Altman is manager of HRIS Analytics and Architecture at Memorial Sloan Kettering Cancer Center. He is responsible for putting actionable data in the hands of workers to assist in decision-making, manages a work stream of their Workday implementation, and determines short and long-term application architecture strategy. Previously, he was founder/CEO of Peopleserv, a software/services company. He has published extensively and serves on IHRIM Workforce Solutions Review editorial committee and can be reached at Jeff Higgins, Contributing Editor Jeff Higgins is the CEO of the Human Capital Management Institute, a driving force in workforce analytics helping companies transform data into intelligence via workforce planning and predictive analytics. Higgins is a founding member of the Workforce Intelligence Consortium, a member of the ISO Technical Advisory Group (TAG) developing human capital standards, board member of the Center for Talent Reporting (CTR) and editorial committee member for IHRIM’s Workforce Solutions Review (WSR) magazine. He can be reached at

In this issue of WSR, we explore this new and expanding field of artificial intelligence (AI) from a number of different vantage points. AI has significant implications for the way work is accomplished in today’s organization. Not only does it have the power to improve efficiency and decision-making, but it creates a wide range of new services and capabilities that have previously not existed. These changes will have an impact on how we think about the workforce, the skills and capabilities they will need to be successful in the future, how they will be organized, and how they will work in tandem with machines that can learn independently and provide guidance on a scale not previously seen. At the same time, the HR function itself is at the nascent stage of applying AI technology to its own processes. From recruiting, to learning and development, to traditional HR operations, artificial intelligence offers the promise of reducing the amount of repetitive work that often plagues HR professionals, while providing them and the employees they enable with more effective guidance and support. The first set of articles, “What to Do When Machines Do Everything” by Ben Pring and “AI and Shared Services Practical Applications in Business” by Shel Holtz, provide an overview of AI and the recent progress companies have made with the various technologies that fit under this umbrella. They demonstrate the potential opportunities these new technologies offer across a number of different industries, from automotive to financial services to healthcare. Both articles describe how AI is rapidly moving out of the laboratory and into the core business offerings of many firms. Lior Gazit’s article, “Trading Cancer for Data: Machine Learning for Cancer Diagnosis” takes a detailed look at how machine learning, pattern recognition, and voice recognition are being applied in the field of medical diagnostics. Using research conducted at Memorial Sloan Kettering Cancer Center, Gazit describes a real-life application of how artificial intelligence can be applied to the interpretation of medical imaging, and the implications for other types of diagnostic approaches. He also highlights the importance of artificial intelligence in augmenting, rather than replacing, human judgement and expertise. The next two articles by Roy Altman and Al Adamsen discuss some of the potential implications for the use of AI in the workplace. They explore not only how AI will work together with other technologies, but will require companies to consider a host of items, including how employees will co-exist with machines in making decisions, the legal, social and educational implications for AI, and how these various technologies will affect the future employee experience. John Sumser’s article focuses on the ethical issues associated with AI technology. He highlights the need for companies to better understand who has ultimate responsibility for the decisions made using AI, how are those decisions carried out within an organization, and how you address biases in existing data. As AI technology use expands, both users and software providers will need to tackle these ongoing challenges. The remaining articles focus specifically on the impact of AI on the human resources function. Janet Merten’s article highlights one of the initial studies focusing specifically on how cognitive technologies will have an impact on HR. The article identifies how senior executives think artificial intelligence will benefit the HR organization, using insights from CEOs and CHROs. Her study poses an important question, “Will employees take guidance from cognitive systems, and if so, what kinds of advice?” As we have seen with new technology introduction, the battle for success is won or lost by the willingness of the end user to trust the insights they are getting and use them accordingly. Merten’s article highlights one potential use case for artificial intelligence: the use of “bots” to answer employee questions – one of the traditional mainstays of the HR function. Freddye Silverman’s “Bots ‘R Us” article and Katherine Jones in her Back Story article, “May the Bots be with You…” describe the development of chatbots in different customer-facing situations, and how those technologies might be applicable to the world of human resources. And, our interview with Catharina Lavers Mallet, the head of product development for Talla, provides unique insights and lessons learned from an emerging startup that is rolling out chatbots within the HR function. We hope the articles in this issue give readers a sense not only for the variety of applications that AI will enable in the future, but the potential considerations that individuals need to take to adjust to the new world of augmented insight. • Workforce Solutions Review • July-September 2017



Ben Pring, Cognizant Technology Solutions

What to do when Machines do Everything Artificial intelligence has left the laboratory (and the movie lot) and is in your building. It’s in your home. It’s in your office. It’s pervading all the institutions that drive our global economy. From Alexa to Nest to Siri to Uber to Waze, we are surrounded by smart machines running on incredibly powerful and self-learning software platforms. And this is just the beginning. To date, we’ve been enjoying – without even really noticing – various forms of “weak” artificial intelligence (AI). It’s how Amazon recommends just the right gift. How Netflix suggests the perfect film for your Sunday evening. Or how Facebook fills your newsfeed. These forms of AI have been welcome little helpers, making our days just a bit easier and more fun. Once we start using them, we stop thinking about them. In just a few short years, these machines have become almost invisible to us in our personal lives. Now, AI is transitioning from being our little daily helper to something much more powerful – and disruptive – as the new machines are rapidly outperforming the most talented of us in many endeavors. For example: • Games of intellect – Artificial intelligence platforms can now outcompete us at some of our most challenging games: Jeopardy, Chess and Go. Google’s AlphaGo beat world champion Go player Lee Sedol by a score of 4-1 in March 2016.1 A convincing win, but not a rout. Yet, with the current rate of technological advancement, in just a few years it will be inconceivable for a human to beat the new machines in such games of the mind. •


Driving – The driverless car, while still relatively nascent, is already a better driver than the average person. According to a Virginia Tech study, human-driven vehicles are involved in 4.2 crashes per

July-September 2017 • Workforce Solutions Review •

million miles versus 3.2 crashes per million miles for the automated car.2 This disparity in safety will undoubtedly grow considerably in the next few years, and driverless cars – which never text behind the wheel or drive drunk – may soon become mainstream. •

Trading – In 2015, six of the top eight hedge funds in the U.S. earned around $8 billion based largely – or exclusively – on AI algorithms.3 The machine has already won in stock-picking.

Healthcare – In medicine, the new machine is quickly surpassing the capabilities of human radiologists (editor’s note: see “Trading Cancer for Data: Machine-Learning for Cancer Diagnosis” in this issue). Researchers at Houston Methodist Hospital utilize AI software, which interprets results of breast X-rays 30 times faster than doctors and with 99 percent accuracy. By contrast, mammograms reviewed by humans result in unnecessary biopsies nearly 20 percent of the time.4

We could go on and on with many more examples, but the point is clear: the new machines have already surpassed human capability in many ways. And, with the geometric growth in the power and sophistication of these platforms, this is only a preview of coming attractions. Thus, this rapid expansion of AI leads us to ask some big questions: • Will a robot take my job away? • Will my company be “Ubered?” • What will my industry look like in 10 years? • Will my children be better off than I am?

Like It or Not, This Is Happening

What the World Economic Forum hailed in

2016 as the Fourth Industrial Revolution is now upon us: a time of economic dislocation – when old ways of production give way to new ones, and when those who can harness the power of the new machine will harvest the bounty of economic expansion.5 In the same manner that the first Industrial Revolution was powered by the invention of the loom, the second by the steam engine, and the third by the assembly line, the fourth will be powered by machines that seem to think – what we refer to as “systems of intelligence.” As such, whether you are managing a large enterprise or just starting your first job, deciding what to do about the new machine – this new cocktail of AI, algorithms, bots, and big data – will be the single biggest determinant of your future success.

Digital that Matters

For the past decade, we’ve collectively enjoyed “digital that’s fun.” We’ve seen the incorporation of Twitter (2006), the introduction of Apple’s iPhone (2007), and Facebook’s IPO (2012). These companies – along with others, such as Google, Netflix, and Amazon – have been able to generate unprecedented commercial success in terms of customer adoption, daily usage, and value creation by changing how we communicate and socialize. Yet, history will note that we started the digital revolution with the amusing and the frivolous: Facebook posts, Twitter feeds, Instagram photos. We are using the most powerful innovations since the introduction of alternating current to share cat videos, chat with Aunt Alice, and hashtag political rants. However, that’s just the warm-up act, for we haven’t yet begun to fully realize the potential of the new machines. Technology writer Kara Swisher summed it up best when she said, “In Silicon Valley, there are lots of big minds chasing small ideas.”6 Well, we’re entering an era of big brains focused on big ideas – digital that matters: using these technologies to transform how we are educated, fed, transported, insured, medicated, and governed. While companies like Facebook, Amazon, Netflix, and Google (sometimes known as the FANG vendors) seem to have established themselves as the presumptive and eternal winners in this space, history will likely remember them as the precursors to a much more momentous and democratic economic shift. The next wave of digital titans probably

won’t be characterized by start-ups from Silicon Valley; instead, it will be made up of established companies in more “traditional” industries – in places like Baltimore, Birmingham, Berlin, and Brisbane – that figure out how to leverage their longstanding industry knowledge with the power of new machines. We’re starting to see this play out as we collectively work to apply systems of intelligence to help address some of our most vexing societal ills in areas where digital technology is not just entertaining or convenient, but also life-altering. Certainly, many of our institutions – the pillars of our society and our everyday lives – are ripe for improvement. For example, worldwide we lose 1.2 million lives to car accidents annually, with more than 94 percent of these accidents a result of human error.7 In the U.S. alone, these wrecks cost society over $1 trillion. This is nearly one-third the amount the U.S. federal government collects in individual income taxes.8 Driverless cars promise to save countless lives and heartache. Medical misdiagnoses could also plummet. Right now, 5 to 10 percent of trips to the ER results in a misdiagnosis.9 More than 12 million diagnostic mistakes contribute to 400,000 deaths caused by preventable errors each year… and that’s just in the U.S.10 Applying data to the diagnostic process could dramatically improve patient outcomes. The U.S. spends more per student on secondary education than most other countries in the world, but generates mediocre results. In a recent international study, American students achieved scores far below those in many other advanced industrial nations in science, reading, and math.11 By tailoring lessons to the individual learning style of each student through technology, we could make the education process radically more productive and effective for both students and teachers. These are the sorts of big things that we can address with the new machine. It’s digital with purpose, digital that matters. And, the big brains bringing these innovations forward will not necessarily reside in Silicon Valley or an MIT dorm room. They may well be sitting in an office down the hall at your company. For example, McGraw-Hill Education is applying new technology to help teachers and kids improve learning with a system called ALEKS. The artificially intelligent Assessment and LEarning in Knowledge Spaces system uses adaptive questioning to quickly and accurately

It’s digital with purpose, digital that matters. • Workforce Solutions Review • July-September 2017


About the Author

Ben Pring joined Cognizant in September 2011 and leads The Center for the Future of Work. He is a co-author of the new book What To Do When Machines Do Everything (2017) and the best-selling and award winning book, Code Halos; How the Digital Lives of People, Things, and Organizations are Changing the Rules of Business (2014), and co-developer of the accompanying award winning app. He has authored numerous other white papers on areas such as the economics of big data and the primacy of the digital customer experience. He joined Cognizant from Gartner where he researched and advised on areas such as cloud computing and global sourcing. At Gartner, he was one the of the lead analysts on all things “cloud;” he wrote the industry’s first research notes on cloud computing (in 1997!), on (in 2001) and became well-known in the IT industry for providing predictions of the nature and velocity of the change as cloud computing became the foundation for the next wave of competition in global IT. His expertise took him to Cognizant where his charter is to research and analyze how clients can leverage the incredibly powerful new opportunities that are being created as new technologies make computing power more pervasive, more affordable, and more important than ever before. He has a degree in Philosophy from Manchester University in the UK, where he grew up. He can be reached at


determine exactly what a student knows and doesn’t know in a course. ALEKS then instructs the student on the topics he or she is most ready to learn. As the student works through a course, ALEKS periodically reassesses the student to ensure retention. All of this results in more flexible, one-on-one instruction for students, which boosts student success. And for teachers, ALEKS helps take over some of the more routine – and, let’s say it, boring – work to allow them to focus more intently on working with students. Another example can be seen at Discovery, one of South Africa’s leading insurers, which uses its Vitality platform to provide economic incentives – discounts on travel, entertainment, healthy food, gym memberships, sports equipment, health products, etc. – to its members based on whether they participate in healthy behaviors. Members earn points by logging workouts with connected fitness devices and purchasing healthy food (also logged by swiping their Vitality card). The insurance sector may not be known as a hotbed of innovation, but Discovery has built a thriving business based on the value derived from the new machine. Examples like these are about to be replayed a thousand-fold across all sectors of our economy. So the question becomes: Will you play, or stand on the sidelines?

Will Humans Be Automated Away?

We have already proven that we love to consume AI-based products (with our rabid usage of the FANG vendors’ offers on our smartphones). Yet, once we get over our initial awe of the new machine, we start to wonder how it will impact jobs. What will happen to all those bankers, drivers, radiologists, lawyers, and journalists? What will happen to…me? Will a robot take my job? Many of us don’t know if this Fourth Industrial Revolution is good, bad, or a mix of the two. It all starts to feel like a capitalist’s dream…but a worker’s nightmare. And, the uncertainty is creating a palpable sense of anxiety, for, at a personal level, many of us don’t know what to do about it. Some see only the dark side of this shift and, indeed, many of today’s headlines forecast a grim future in a “jobless economy” as robots take over our livelihoods. But, the coming digital boom and build-out will be highly promising for the prepared. In fact, it will usher in once-in-acentury growth prospects as we re-engineer our infrastructure, our industries, and our institutions. Similar to the prior three industrial revolutions, this one will steamroll those who wait and

July-September 2017 • Workforce Solutions Review •

watch, and will unleash enormous prospects and prosperity for those who learn to harness the new machine. All of this depends on what you do now to prepare for an era when machines can potentially do nearly everything related to knowledge work. Will many jobs be “automated away” in the coming years? Yes! However, for the vast majority of professions, the new machine will actually enhance and protect employment. We don’t think, for example, that a single teacher or nurse will lose their job due to artificial intelligence. Instead, these professions will become more productive, effective … and enjoyable. Additionally, entirely new professions will be created, driving employment in fields we can’t currently envision (imagine trying to describe a “database administrator” to somebody in 1955). We have much to look forward to…if we understand exactly what the new machine can and cannot do, and how it will impact the future of work.

Getting Ahead in the Age of the New Machine

We believe there is a structured approach for moving forward in this age of confusion, peril, and reward. We call it the AHEAD model, which we have developed after working with many Global 2000 companies at the vanguard of digital transition. The acronym stands for: • Automate: Outsource rote, computational work to the new machine. ◦ This is how Netflix automated away the Blockbuster retail store, and how Uber is automating away taxi dispatching. • Halo: Instrument products and people, and leverage the data exhaust they generate through their connected and online behaviors (what we call “code halos”) to create new customer experiences and business models.12 ◦ General Electric and Nike are changing the rules of the game in their industries by instrumenting their products, surrounding them with halos of data, and creating new value propositions and customer intimacy. • Enhance: View the computer as a colleague that can increase your job productivity and satisfaction. ◦ The GPS in your car currently enhances your driving – keeping you on the fastest route, alerting you of road hazards,

and ensuring that you never get lost. In the coming years, entire vocations – from sales to nursing to teaching – will be revolutionized with the power of computer-based enhancement. Abundance: Use the new machine to open up vast new markets by dropping the price point of existing offers, much as Henry Ford did with automobiles. ◦ In the way that Betterment is using AI to bring financial security to the masses, what market offers can be greatly democratized and expanded in your industry? Discovery: Leverage AI to conceive entirely new products, new services – and entirely new industries. ◦ As Edison’s light bulb led to new discoveries – in radio, television, and transistors – today’s new machine will lead to a new generation of discovery and invention.

The first “play” outlined in our model – to automate – is the one most prevalent in today’s zeitgeist. Automation has been the initial step in each industrial revolution, as one loom replaced 40 textile workers, or one steam engine had the power of 50 horses. Today, automation will be

a similar necessary “evil” – for it’s how you will deliver at the “Google price” in core portions of your company. However, what most market observers miss is that the next wave of automation will pave the way for invention and economic expansion through the four subsequent plays. This one-two of efficiency plus invention will manifest itself across all industries, and, as outlined above, much of this shift will not be driven by companies started last year – or even 10 years ago – but by companies started by our grandparents. This is because those companies have access to the richest lodes of data – the “fuel” for the new machine. The title of our new book is What to Do When Machines Do Everything. This may sound a bit hyperbolic, and clearly machines will never do everything – and nobody really wants them to. But, in the next few years, the new machines will be embedded almost everywhere and in almost everything, and will increasingly do more and more of the work people do today.​ Technology is no longer the domain of the few, but the province of the many. As such, those who win in the next phase of the digital economy are not those necessarily who can create the new machines, but those who figure out what to do with them. That means you.


Christopher Moyer, “How Google’s AlphaGo Beat a Go World Champion,” The Atlantic, March 28, 2016, http://www. 1

“Automated Vehicle Crash Rate Comparison Using Naturalistic Data,” January 8, 2016, featured/?p=422. 2

Emel Akan, “World’s Top Hedge Fund Managers Took Home $13 Billion in 2015,” Epoch Times, May 17, 2016, http://www. 3

Todd Ackerman, “Houston invention: Artificial Intelligence to read mammograms,” San Antonio Express-News, September 16, 2016, 4

Klaus Schwab, The Fourth Industrial Revolution, World Economic Forum, Jan. 11, 2016, 5

John Kennedy, “Kara Swisher: ‘In Silicon Valley, There Are a Lot of Big Minds Chasing Small Ideas,’” Silicon Republic, June 24, 2015,\. 6

“Human Error Accounts For 90% of Road Accidents,” International News, April 2011, fleet-alert-magazine/international/human-error-accounts-90-road-accidents. 7

See and publications/road_traffic/world_report/en/ and 8

“Surprising Number Of Emergency Room Medical Errors,” July 15, 2016, surprising-number-of-emergency-room-medical-errors/. 9 and http://www. 10 and fact-tank/2015/02/02/u-s-students-improving-slowly-in-math-and-science-but-still-lagging-internationally/. 11


For more information on Code Halos, see our white paper and book, • Workforce Solutions Review • July-September 2017



Shel Holtz, Holtz Communication + Technology

AI and Shared Services Practical Applications in Business After years of discussion (and science fiction), artificial intelligence (AI) is upon us, and while it is still fairly nascent, developments are happening at a frantic pace. The biggest players have familiar names: Google, Facebook, Amazon, Apple, and IBM. Some of the earliest implementations demonstrate AI’s potential for disruption. Consider, for example:


The Associated Press is using AI to write hundreds of stories summarizing minor league baseball games. No reporters are being displaced; AP never had enough reporters to cover minor league teams, but because stats are available for every game, the AI can parse the numbers and assemble a story after reading thousands of game recaps in order to learn the nuts and bolts of a sports article.

An AI “watched” thousands of horror movie trailers, then “watched” an entire horror movie. Having learned the elements of a horror movie trailer, the AI crafted its own trailer.

On Pinterest, when users find an item they like (a couch, for example), the AI will find similar couches based on the visual characteristics. Google Autodraw allows users to draw a picture on a touchscreen. The AI is able to figure out what the user is drawing and dish up some finished sketches. Google Lens, announced at the Google developer’s conference this year, will not only let your smartphone understand what it sees (the kind of flower you’re looking at, for instance), but also take action (connect to WiFi after seeing the network name and password).

Adobe has developed software that can replicate any human’s voice so you can have that individual “say” anything. All the software needs is 20 minutes worth

July-September 2017 • Workforce Solutions Review •

of that person’s voice. Another program can be shown an object (a volcano or a bird) and produce original photo-quality images that are similar to the one it was shown. In one experiment, all you need to do is give the AI a caption (such as “a pitcher is about to throw the ball to the batter”) and the AI can create original photo-quality images of just that. Artificial intelligence – also known as “cognitive computing” – learns based on its experiences, which has some people worried. Voices like Bill Gates and Stephen Hawking have warned of the risks of AI, which range from unethical uses to the Skynet scenario from the “Terminator” movies, in which a fully evolved, sentient AI has no further use for humans. (To address ethical concerns, the big players have formed an alliance, Partnership on AI, which will “study and formulate best practices on AI technologies, advance the public’s understanding of AI, and serve as an open platform for discussion and engagement about AI and its influences on people and society.”) When it comes to the workplace, though, the uses of AI are far more mundane than science fiction writers envision. In some cases, they can be troubling. Fukoku Mutual Life Insurance, a Japanese company, has laid off 30 employees, replacing them with IBM’s Watson AI system. The company expects Watson to be 30 percent more productive than those employees. Fukoku will recoup its investment after only two years while saving the company well over US$1 million each year. IBM claims Watson can “analyze and interpret all of your data, including unstructured text, images, audio, and video,” according to an article in The Guardian. Fukoku’s AI is expected to consume volumes of medical documents and weigh other information in order to determine how much to pay on claims, even though the final amount will still be approved by a human staff member.

The Fukoku tale is one example of how AI can affect shared services in just about any organization. In some cases, AI could (as many have predicted) replace workers. In most, though, it will assist and support workers. As IBM’s senior vice president of communication and marketing, Jon Iwata puts it, AI should stand for “augmented intelligence” instead of “artificial intelligence.” To understand how AI will complement company staff, it’s important to understand exactly what AI is and how it works. Simply stated, AI computer systems perform tasks that normally require human intelligence. That doesn’t mean they think the same way people do. Instead, they consume vast amounts of information – both structured data (like a database) and unstructured data (emails, for example, or books). When presented with a question or problem, AI systems review and sort the data, find the information that’s relevant, and assign a probability score to the best answer. (If the probability is under 50 percent, most AI systems won’t answer.) Artificial intelligence systems are able to perform all these tasks in seconds, which is how IBM’s Watson was able to beat the two winningest Jeopardy champions. As you might expect, AI systems are good at predicting. They can predict the best chess move (which is how IBM’s Big Blue beat world chess champion Garry Kasparov) and the best Go move (resulting in Google’s AI beating the world champion Go player 4-1). A university-developed AI system called DeepStack has beaten world champion poker players, a particularly impressive feat in that, unlike chess and Go, the AI can’t see the whole board and anticipate every possible move. All the cards haven’t been shown and a player may opt to hold or not place a bet. That’s where machine learning comes in, according to an article in Science magazine: “The program continuously recalculates its algorithms as new information is acquired. When the AI needs to act before the opponent makes a bet or holds and does not receive new information, deep learning steps in. Neural networks, the systems that enact the knowledge acquired by deep learning, can help limit the potential situations factored by the algorithms because they have been trained on the behavior in the game. This makes the AI’s reaction both faster and more accurate.” While AI is incredibly efficient at prediction based on its unparalleled ability to process oceans of data and learn as it gains more

experience and new data, it’s not so great at making judgments. That’s where humans come in. While some experts worry that AI will replace nurses – they’ll do a better job of analyzing symptoms and predicting which treatment will produce the best outcome – it’s more likely that the job of nursing will change. Human nurses will use AI to identify the best treatment but will use their own judgment and humanity to deliver that care. AI will most likely play the same role in business shared services. A public relations department routinely drafts press releases to review the financials for the last quarter. These releases are fairly routine and based almost entirely on data. It’s not hard to imagine an AI system that has reviewed thousands of quarterly earnings releases getting the assignment to create this quarter’s release using current numbers. The AI could even review the release to make sure it complies with rules and regulations from the Securities and Exchange Commission and other regulatory bodies. Of course, no company would transmit a release to investors directly from an AI. Somebody needs to get the right quote from the CEO, for example, and make judgment calls on the tone the release employs based on the company’s performance. Marketing is another example of a shared service that will be able to exploit AI for huge advantage. In April of this year, Nielsen announced the introduction of Nielsen Artificial Intelligence. This adaptive learning technology is part of the Nielsen Marketing Cloud and, according to the press release, will enable “clients to respond instantly to real-time changes in consumer behavior, resulting in more relevant content and advertising, higher levels of customer engagement, and improved ROI.” Artificial intelligence is bound to find its way into virtually all the activities of a Human Resources department. IBM is rolling out a Watson-based system called IRIS, which prioritizes open positions, examines market dynamics and company history to identify the most difficultto-fill positions and which need to be filled most quickly. It then analyzes applicants’ cover letters, résumés, and job histories. And when it comes to marketing job openings, IRIS even compares the company’s Glassdoor reviews and media coverage to industry competitors. There are already a number of AI-driven chatbots that engage with job candidates, answering routine questions and soliciting information and feedback from candi-

To understand how AI will complement company staff, it’s important to understand exactly what AI is and how it works. • Workforce Solutions Review • July-September 2017


About the Author

Shel Holtz is a speaker, writer, consultant, trainer, and podcaster focused on organizational communication, particularly as it is affected by emerging technologies. Before launching his own business in 1996, he worked as director of corporate communications in two Fortune 500 companies and as a communications consultant in two global human resources consulting firms. He has written six communicationthemed books and countless articles. Holtz is a Fellow of the International Association of Business Communicators, a Founding Fellow of the Society for New Communication Research, and a Platinum Fellow of the Mayo Clinic Social Media Network. You can find him online at


dates. Artificial intelligence can also crunch the data about the candidates for a job and predict which one will wind up being the best employee. The same is true for identifying which employees deserve promotion or will make the best managers. Human resources experts anticipate AI playing a role in keeping new hires informed and engaged between the time an offer is made and the employee’s first day on the job. Artificial intelligence will also help with onboarding, answering questions, and delivering timely updates. Again, it’s important to remember that AI will handle routine, mundane tasks. As a post to the IBM Smarter Workforce blog points out, “The importance of the human element remains. Human Resources is going to have to continue to focus on the employee experience – and how an organization treats its employees throughout the entire life cycle, allowing AI to assist with the more laborious tasks.” Chatbots – referenced above – are already emerging as one of the early leading uses of AI. Chatbots are computer programs that emulate human interaction. It’s likely you have talked with a chatbot when getting customer support online or even on the phone. While some chatbots are simple programs that only respond to specific queries, AI increasingly is in play. Systems from Microsoft, Facebook, and others let virtually anyone create a chatbot that understands natural language inquiries and delivers relevant, meaningful answers in equally natural language. Artificial intelligence-enabled chatbots are already making their presence felt in the enterprise. At, a chatbot now handles call center employees calling in sick. Instead of calling a hotline and talking to a human being, employees use their phones to text a message to Mila, who can notify supervisors and update the employee on the number of sick days they have remaining (and sign off by hoping the employee feels better soon). Jane is an AI-fueled chatbot from Bobby Mukherjee, a Silicon Valley entrepreneur, that answers basic HR questions via chat apps like Facebooks’ Messenger. It can also analyze questions that employees ask, revealing potential problems with company policies. Then, there’s TangoWork, an employee communication chatbot platform that runs

July-September 2017 • Workforce Solutions Review •

on Microsoft’s foundation, making it possible for employees to interact regardless of their messaging app of choice (Messenger, WhatsApp, Kik, Kako, Line, or any of the others that are especially popular with GenZ employees). Employees can ask about today’s news and TangoWork will deliver the top headlines. Employees then select the article they want to read; TangoWork delivers the meat of the story and then offers options about which angle the employee wants to learn about next. TangoWork can also update employees about shift changes and other timely news. Artificial intelligence chatbots are also frontand-center in the exploding field of voice tech, which (so far) Amazon owns with its Alexa series of appliances (the Echo, Tap, and Dot). Google, Samsung, and Apple are among the other players entering the field. It won’t be long before employees are speaking their queries into invisible technology and getting responses in a friendly, natural voice. If you have an Echo, you probably already ask for the news and Alexa delivers recorded updates from the services you have selected, from National Public Radio to The Washington Post. While we’re not dealing with virtual, augmented, or mixed reality here, it’s worth noting that the enterprise is already using these technologies. They all will employ artificial intelligence for speech and image recognition. All of these shared-service, AI applications just scratch the surface of how companies will improve efficiency and productivity, while reducing costs. The companies developing AI are only at the beginning of their journey. As the technology evolves, new uses will emerge at a dizzying pace. At the same time, AI will displace some jobs and companies will need to address these layoffs with humanity and compassion. Some organizations will undoubtedly endorse broad societal solutions to the reductions, such as universal basic income, which could tax AI and robots in order to provide displaced workers with money to live on while exploring other possibilities. The one thing shared service managers cannot do is ignore AI. As competitors adopt it, companies will either fall in line or fall by the wayside.


Lior Gazit, Memorial Sloan-Kettering Cancer Center

Trading Cancer for Data: Machine Learning for Cancer Diagnosis Introduction

Artificial intelligence (AI) is penetrating everything we consume, create, and own. This is an industrial revolution that has been expanding for decades and will continue until it reins in every bit of redundant human labor. The first industrial revolution, starting in the 1700s, took manufacturing out of people’s homes and into coordinated factories, setting the ground for efficiency and automation. Machines were the key. A worker with the aid of a machine introduced an unimaginable level of production. The second revolution, which introduced electrification, was based on the premise that the worker was just slowing the machine down. A machine with the aid of electrification introduced a significantly increased level of production. First, we had machines empower the work, then we designed them to precisely do the work, and now we are teaching them how to learn to do the work; they learn it, and then do it. These machines do the work better than we do. A stock trading algorithm replaces stock brokers, plural. It is created by a programmer who typically has a great understanding of logic and decision-making, but probably did not even take “Introduction to Microeconomics.” It is then taking years’ worth of financial data to derive optimal trading rules. A typical algorithmic trading (algo-trading) program may perform millions of trades a day while it also reads the news, adjusts to changes in the market, and spouts out updates to management. Power consumption is optimal, space consumption is minimal, the program doesn’t take vacation days, and is oblivious to ego. If a human can be trained to spot a preferable trade, so can a machine. What else can machines be trained to do? Solve a mathematical equation, hit an enemy target, tell when to flush, signal

before hitting a parked car’s bumper? Simple enough! Recognize John Doe’s face out of 10 hours of surveillance footage, drive a car, and spot a cancerous cell in an MRI?

Artificial Intelligence

Artificial intelligence is defined as programmed decision-making. As this is a very broad definition, some argue that the label AI only applies to highly sophisticated applications, thus adding a contemporary aspect to it. Spellcheck, for instance, is a solution to a simple problem. When a word is entered, the program searches for it in the dictionary, and if it isn’t found, it is flagged as a potential error for the user to correct. Autocomplete will take a partially entered sentence, search past entered sequences that started with identical text, and complete the entered text in the most probable (frequent) manner. Sentiment Analysis observes the text and derives what attitude the user is expressing, utilizing advanced algorithms in Natural Language Processing (NLP). Are they upset, disappointed, happy, confused? The answer will surely come in handy when trying to automatically prioritize the 10,000 customer service queries submitted in the last 24 hours, before the morning shift starts hitting the phones and keyboards. Spellcheck is a remarkably simple algorithm. Given a preset dictionary, the decision for each word is deterministic and simple. With Autocomplete, the decision is statistical, there may be two completions that are plausible, but one appeared 1,000 times in the past, while the other appeared 50,000 times, thus deeming the latter more likely. The algorithm uses past entries to learn the probability of sequences and then applies the principle of Maximum Likelihood by selecting the most frequent. It learned • Workforce Solutions Review • July-September 2017


on its own, no prior insight was programmed into it by the programmer. This entire notion is called machine learning (ML). The program uses past instances to learn about recurring patterns, an empirical study of cause and effect. So far so good! With Sentiment Analysis, we are entering fresh territory. The machine attempts to take a text sequence that it has probably never seen before, and classify it as expressing a specific attitude or emotion. It must break the text down to individual words, as we humans do, and key on those words that express opinions, emotions, moods, etc. Is the mere presence of the sequence “…thrilled when I opened the package…” enough to tell the machine whether this is a positive or negative query? Not quite. It must analyze the rest of the statement to validate. It is enough for that sequence to be preceded by “not” to turn it on its head. This calls for a thick sauce of analyzing keywords and their interactions while utilizing statistical learning of past examples so as to understand how this specific mixture comes together, and hopefully derive what attitude it is suggesting.

Machine Learning

Figure 1. Four examples of typical images from a dashboard camera. When driving, each would be analyzed by the recognition algorithm to identify stop signs.


Let’s focus on machine learning. As conveyed in the previous section, machine learning drives the current AI wave we are experiencing. It is the force multiplier behind this new revolution. No longer is technology limited to what humans can perceive. Instead of specifying a set of rules for what action to take, the rules are set for

July-September 2017 • Workforce Solutions Review •

how to learn. Inherently, the rules for action immediately follow. There is a price, though. You see, in the first industrial revolution one would have to provide space for the workers to come and coordinate their work – factories. With the second, one would have to go through the trouble of building an entire assembly line, only after which it would be put into work, which is a big investment. With machine learning, something has to make up for the fact that you don’t specify the solution, but are forcing the machine to learn. It needs something to learn from: data. I’d like to welcome you to any ordinary graduate class in computer vision (CV). We need to design an algorithm for a digital camera that identifies stop signs in images. This would be a part of an AI of an autonomous car. The code’s input is a single image at a time, and the output for each image is either “Stop Sign Ahead,” or “Safe to Go.” Assuming we are taking a course in computer vision before taking our first class in machine learning, we would probably choose to teach the algorithm “what a stop sign looks like.” You will notice, as depicted in Figure 2, for the machine, an image is a set of ordered numbers (pixel values), thus the analyses are limited to looking for sequences of values in the ordered structure that we portray as an image. So, to mine for a stop sign, we will define rules. It will be a long process filled with analyses of the image and questions that follow. We may scan the image for straight lines. We would check if any eight of them form an octagon. If so, we would try to discriminate it farther so to avoid calling just any octagon a stop sign: Are most of its inner pixels red? Are the inner pixels that aren’t red, white? This would be a long piece of code. Eventually, assuming we missed nothing, we would have a solid algorithm. In today’s era, we are not thrilled about such approaches. Two main caveats come to mind: R&D time and intellectual limitation. How much time would it take us to design the stop sign algorithm? Don’t be fooled into accounting only for the time it took us to set the rules for recognizing stop signs, we had actually spent quite a substantial amount of time in our lives learning what stop signs look like to begin with. To make this point clearer, what if the task was to recognize something less trivial? For instance, design an algorithm that takes a magnetic resonance image (MRI) of the spine and tells whether the subject suffers from a ruptured

disk. If we went with the above approach of setting specific rules, our first step would be to start studying MRIs of the spine so to understand what our algorithm must look for. What do healthy spines look like? What do ruptured disks look like? Only after we grasped the details, would we be able to articulate them and suggest a set of concrete rules. We would probably spend a few weeks before we had something reasonable to start with. Now let’s discuss our own intellectual limitation. While we wouldn’t have a hard time suggesting a set of rules for stop signs, how about for dogs? Recognizing a dog in a clear image is an unremarkably easy task for a human, even if their brain hasn’t grown completely, such as in young children. But, could we specify a set of rules that would apply to an image and its pixels to identify a dog? Oh man! We would probably not get past the “Does it have four legs?” question. While we do have the intellectual ability to recognize a dog, we can’t articulate a set of (pixel-based) rules that would apply to an image’s pixels. These aspects and challenges are not limited to computer vision problems. They exist in every machine-learning endeavor. If, for instance, LinkedIn or Facebook wants to suggest content or potential connections to their members, they would face similar problems. They would want to make a match based on a member’s posts, interests, etc. But, articulating rules for how a member’s posts relate to other content, what they may find interesting is as complicated as the examples of the dog and the spine. Machines to the rescue! A machine-learning algorithm will do the job. Quickly too! How would it work? Since I started with the notion of computer vision, I’ll stick with it. We would provide the machine with a big enough variety of images. How big? It needs to be big enough for it to get an idea of what a positive input looks like, and a negative one. Much like we would research for ourselves to learn what a ruptured disk looks like, we would have two sets of images to train on, those labeled a “Yes” and those labeled a “No.” In machine learning, this is called a labeled data set. Next, a specific machine learning algorithm of the developer’s choice will iterate over all the “Yes” images and will find patternlike consistencies between the pixels. It will then do the same for the “No” images. It will see what consistencies exist in the “Yes” images, but don’t also exist in the “No” images. If it finds such

consistencies, they become the rules for classification. Voila! Remember the ruptured disk and the R&D time? Well, what would take a human weeks, a machine can do in mere seconds. Remember the dog and our intellectual limitation to articulate pixel-based rules? The machine does not learn first what a dog is and then attempt to explain it in pixel form (like a human would); it directly looks for pixel patterns that are typical of dogs. C’est tout!

Figure 2. A look into Computer Vision. On the right is what humans see; on the left is what the algorithm “sees.” It perceives the image as an array of values. Each pixel has a position and a value. Note that the values in the right image are of the intensity of the brightness, presented for simplicity. In practice, each pixel has not one but three values, typically one for each of the three colors, red, green and blue.

My Pancreas Work

As a data scientist on the Strategy Analytics team in Memorial Sloan-Kettering Cancer Center, I’m like a kid at Six Flags. We utilize some of the most cutting-edge algorithms and we deal with a lot of data, usually two kinds: clinical and operational. Clinical data would be on the patient level, featuring disease condition, physical characteristics, medical history, etc. Operational data would present hospital processes like occupancy, flow of chemotherapy administration, and so on. A typical project would result in a prediction or classification algorithm. The purpose of the algorithm’s result may be operational, and the data used is a mix of operational and clinical attributes. In Strategy Analytics, we don’t study cancer (what I typically call the “micro” of cancer), we learn about cancer and use the knowledge for a bigger scope such as treatment operations (I call it the “macro” of cancer). We look for connections between hormones, cancer stages, radiation therapy, tumor size, even finance, admission time, nursing scheduling, and more. These connections, often hidden and non-trivial, are exploited to learn and improve. As a break from routine, we are encouraged to think about ideas of our own. Given my interest in applying machine learning to computer vision applications, I reached out to a researcher • Workforce Solutions Review • July-September 2017


The future is happening and machines are leading the way for us. They tell us what to buy, how to get to a destination, what to read, and who to watch.

in our institution, Dr. Amber Simpson, and allocated a certain amount of time to work with her team on identifying early risk of pancreatic cancer using medical imaging. Pancreatic cancer is notorious for its high mortality rate, mostly due to late discovery. Computer vision to the rescue! We followed a research paper that spoke about how one could identify discolorations in cysts in the pancreas. We figured, if a human can understand how to find these discolorations, so can a machine…but faster. We first started by collecting images of the pancreas from various patients, then creating a labeled set by letting experts diagnose whether each person is a “Yes” or a “No” for pancreatic cancer risk. Next we made a mixture of the two AI approaches that I described above – self-defined rules, and machine learning-derived rules. The first imitated what the mentioned paper taught us. We told the machine to look at a cluster of pixels that are brighter than the surrounding ones. Then, given these bright clusters, we used machine learning to find what clusters are typical of “Yes” labels, and which are typical of “No” labels. This algorithm produced results that supported the initial paper, thus functioning as a non-invasive contribution to early pancreatic cancer detection.1

Additional Examples

I have touched on NLP and computer vision. The typical third musketeer is voice recognition. One example application is the authentication of the speaker for security purposes. Some banks create a layer of security for their automated phone services by asking the caller to recite a predetermined line. The algorithm crosses the fresh recording with previous acquaintance with the account’s owner and determines whether they are indeed the current caller. Seizure detection in epileptic patients is a non-invasive way to alert others of an occurring seizure. Such seizures can occur without warning and their consequences may be physically harmful to the patient, and, in extreme cases, cause death. While the prevention of such seizures is still unsolved, there is great value in alerting nearby caregivers, for instance, when the patient is at home, in bed sleeping, and naturally isn’t being watched. The noninvasive method is driven by machine learning. A set of sensors are constantly collecting electroencephalogram (EEG) signals through the scalp, and when a pre-learned discriminating


July-September 2017 • Workforce Solutions Review •

pattern occurs, it alerts a caregiver. Recommendation systems and targeted advertising are two applications that are very similar in their AI back ends. An on-demand TV content provider will utilize a recommendation system for learning the interest of every individual viewer and then suggest the content that this viewer would appreciate. The learning is based on this viewer’s past choices. Every watched-content is a “Yes” label in the labeled data set, which is continually growing. The next step is to derive characteristics that this content falls under, and recommend to the viewer similar new content. A characteristic, or rule, could be “Thriller, released in the last five years, starring a female Oscar nominee.” It is highly valuable for the provider to be able to match the relevant part of its content to each individual. Targeted advertising is similar. It would track your purchases online and will recommend the next product you are likely to want. One way to do that is to follow the things you bought recently, look for other people who bought the same things in the past, see what those people bought next, and recommend to you that next product. This is very similar to Autocomplete, right? You would order paint and it would recommend a brush. Order a bathing suit and it would recommend sunscreen. Order a wing suit for base-jumping and it would recommend a probate lawyer.

Computer Vision in Medical Imaging for Cancer Diagnosis When I’m discussing the various aspects and applications of AI and machine learning, I get excited. It is the forefront of an ongoing industrial and social revolution. It is evolving so rapidly that within each decade new social and technological norms are formed and we must either keep up or risk being left behind. Since I’m in the field of healthcare, and cancer, in particular, I mark these realms as the “next to benefit” from this revolution. Healthcare is no sand box; it is very conservative, and for the right reasons. We had to let many other sectors experiment with AI and prove it ripe before applying its force to healthcare. As a preliminary step, it is safest to apply AI as an aid, a force multiplayer, serving the physician, nurse, technician, or other healthcare professional. An even safer approach is to start with diagnosis, rather than treatment.

Cancer imaging diagnosis is being transformed by the AI revolution. The need is very clear: better diagnoses will dramatically reduce cancer deaths. The most effective way to cure cancer is to catch it in its early stage, much like a cavity in your tooth. Machines will learn massive image sets and extract pixel-based rules to provide radiologists with an uncanny ability to generate many more diagnoses with better precision. The diagnostic ability of AI does not fully overlap with a radiologist’s, thus yielding wonderful synergies. It has advantages that a radiologist wouldn’t have, but it lacks critical insights the radiologist possesses; much like a pilot and his autopilot. For example, AI can learn texture patterns that are typical of cancerous cells by focusing specifically on the differences between textures in healthy cells and in cancerous ones. A radiologist may miss the fine, pixel-level details. Moreover, radiologists may only take in so much information when observing the patterns, as they are only human. An advantage that the radiologist has is the qualitative understanding and clinical meanings of what is in the image. It also translates to the ability to extrapolate and make sense of something they are seeing for the first time. This is not the case with AI. In most machinelearning cases, if some notion was not presented clearly and frequently in the labeled data set, AI will not understand where it stemmed from and whether it is indicative of cancer. It learns from examples. The outcome of this specific revolution will be spectacular. Cancer screening will be quicker and more common, which means we will all have easier and cheaper access to cancer-targeted imaging services. Monitoring cancer health will be as common as maintaining dental health. Nowadays, we take for granted our ability to avoid a root canal by having twice-a-year checkups at the dentist. Monitoring cancer

health will be similar to that process. Early stages will be the new cavity and later stages will be the new root canal.

Conclusion The future is happening and machines are leading the way for us. They tell us what to buy, how to get to a destination, what to read, and who to watch. This is a beautiful manifestation of our ongoing yearning to improve and excel. There is no dark side to this. Machines will not turn on us, replace us here on earth, or rule us. They are our force multiplier. While it’s easy to get lost in this AI tsunami, remember that machines are only doing what we design them to do. In principle, this revolution will not do something that the past two revolutions didn’t do. Yes, it will require adapting and learning. Yes, it will make some services and professions obsolete, thus creating changes in the job markets, forcing some to adjust while others will become irrelevant. Take librarians for instance. As the library’s documentation of its inventory went digital, the librarian had to adapt and get familiarized with the new computer-based interface. Things are quite similar for the travel agent, too. Today’s youth will have a lot to think about when they choose a profession. Choosing to be a mail carrier 60 years ago is one thing, choosing it 20 years ago is another. What would you think about an 18-year-old banking on being a mail carrier today? These are opportunistic times and the means necessary to seize this opportunity do not discriminate. You don’t need to belong to a specific gender or race, and your physical strength is irrelevant. This revolution is for everyone. Everyone can read, learn, and adapt. It will serve us better, keep us safer, and make us healthier. Viva la revolución!

Endnotes 1

Lior Gazit, Jayasree Chakraborty, Liana Langdon-Embry, Richard K. G. Do, Amber L. Simpson, Quantification of CT images for the classification of high- and low-risk pancreatic cysts ( aspx?articleid=2608967).

About the Author

Lior Gazit is a data scientist in Strategy-Analytics at Memorial Sloan-Kettering Cancer Center, a world leader in cancer care, research and education. The Strategy-Analytics team is a driving force utilizing state-of-the-art analytical tools and algorithms, and was honored the 2012 INFORMS prize. He designs and carries innovational operations driven by predictive modeling and machine-learning programing. He was previously the senior algorithms developer for the Active Noise Control groundbreaker Silentium. He holds a BSc and an MSc in computer engineering and has authored several scientific publications. He is available at • Workforce Solutions Review • July-September 2017



Roy Altman, Memorial Sloan Kettering Cancer Center

Managing a Human/Robotic Workforce – It’s Closer than You Think! Although the title of this article conjures up images of people and robots working side-byside on a futuristic factory floor, it’s really about how to manage rapidly advancing technology. We currently manage people, both employees and contingent workers, and technology and the vendors that provide it. But, there is a significant difference now: artificial intelligence (AI) pertains to systems that learn based on experience, rather than being programmed to perform specific tasks. This requires a different kind of management. Technology is advancing at exponential rates, and very powerful technologies will be available in a relatively short time frame. By being at the forefront of cultural, business, and technology trends, we can prepare ourselves for the challenges that lie ahead. By robots, I mean any intelligent machines, not just robots that move. Although robots that move are important in many applications such as manufacturing, mining, or vacuuming, the great majority of intelligent machines will be those that “think,” either to assist people or autonomously make decisions.

Managing People

Let’s start this exploration with the assets we’re most used to managing: people. Although people management is as old as work itself, relatively recent changes in the business environment are necessitating re-examination of our organizational structures. As managers, we’re used to managing in a hierarchical organizational model, which means that every level of supervisor is responsible for directing and monitoring the work of those below on the org chart. But the nature of the organization is changing due to the dynamic needs of business. Contingent workers, consultants, and contractors are playing a vital role in the workforce, providing special expertise and additional worker power on specific projects while allowing the company


July-September 2017 • Workforce Solutions Review •

to retain staffing flexibility. In the early 2000s, HR systems began to reflect that trend with the “people” model, whereby employees and contingents can intermingle in HR systems. We are trending toward a gig economy whereby non-employees will perform more and more functions. The numbers of actual employees will shrink as the pool and accessibility of a skilled temporary labor force grows. Relationships with contingent workers will be critical to the strategic plans of companies looking to optimize the new model. Aspects of HR, talent acquisition, vendor management and customer relationship management systems will merge into one, as all worker relationships need to be managed from one centralized repository, both inside and outside the organization. As we’ll see in the next section, managing relationships with vendors will play a crucial role and automation will perform more tasks.

Managing Technology

With the move to the cloud, your vendor is replacing the functionality of your IT staff; being responsible for keeping the software available, performing, and secure while applying more frequent updates. This trend is likely to continue as more of the systems we use move to the cloud. Therefore, the relationship with the software vendor should be baked into your management structures. As machine-learning systems rise in capabilities and are more widely used, there will be a great demand for highly skilled workers, such as data scientists. It will be difficult and expensive for each company to acquire their own data scientists and develop their own systems. This expertise will be embedded in the software we use. The trend toward cloud vendors will continue, as companies with economies of scale can provide the most and quickest innovation.

Relationships with vendors will take on an even greater role in the organizational structure as they become integral partners in our business automation strategy. Therefore, not only is our approach to managing people changing, but our method of managing technology, including the vendor relationship as a critical aspect of the workforce, is changing as well. This will prepare us for the powerful technologies of the present or near future.

The early promise of AI went unfulfilled, which initially gave AI a bad name. But over the decades, AI has crept into many of the systems and processes we now take for granted. This has come about through a few computing trends: •

Rapid advances in computing power – Moore’s Law observed that computing power roughly doubles every two years. This has held true throughout several decades of the computing age. Advanced technology requires a great deal of computing power that wasn’t available a few decades ago.

A preponderance of data – The web has increased the amount of data available at an exponential rate. In subsequent paragraphs we’ll see why this is significant in the creation of intelligent systems.

Advances in programming techniques and advanced analytics – Machine-Learning algorithms are moving away from transactional processing and more towards pattern-recognition.

Types of Automation •

Administrative – The systems of the last 30 years or so have been primarily focused on automating repetitive tasks and transaction processing, which is just the thing computers are good at. Decision-support systems – Some systems these days contain knowledge bases to provide needed and timely information to assist in decision-making. These tools are more oriented as an aid to knowledge-workers, as opposed to operational processing.

Intelligent – Decision-support systems are evolving into intelligent assistants, which use voice recognition to interact with people. Siri and Alexa are early examples of this. However, they will increase in sophistication rapidly, to the point where they anticipate our needs and provide them proactively.

Autonomous – Artificial intelligence, until now, has focused on aiding humans in their business, but we are developing automated capabilities where the machines can manage the work without human intervention, and with greater accuracy.

Sentient – In the not-too-distant future, computing capability will rival humans. Systems will progress to have semantic understanding, which will appear as if approaching consciousness.

Evolution of Technology

Artificial intelligence can be broadly defined as any system that makes decisions. It has been around in concept for decades. Back in the 1980s, we saw what was then called “expert systems,” which consisted of relatively complex decision trees. For you programmer types out there, it amounted to many “if…then…else” statements.

Up until recently, software focused on automating administrative, repetitive tasks. Just as mechanized tools of the industrial age automated manual labor, the early part of the computing age did the “heavy lifting” of calculations and predictable tasks. To get a better sense of this, it’s helpful to understand a bit about computers, and how they differ from the way humans think. Almost all computers we use are designed using what’s become known as the “Von Neumann” architecture (named after computing pioneer John Von Neumann). The central processing unit (CPU), which is often referred to as the “brain” of the computer, essentially does one thing at a time very quickly. This differs greatly from the architecture of the human brain, which consists of about 100 billion interconnected neurons. Each neuron can be thought of as an individual CPU – but there are 100 billion of them! Since neural connections are biological, they work at a much slower speed than electronics. So, as opposed to computers that do one thing at a time very fast, the human brain does billions of things at once very slowly. This is called a “massively parallel” architecture, and is more geared toward pattern recognition, which is something humans do well, but computers haven’t excelled at…yet. While we still don’t know the “software” of

As machinelearning systems rise in capabilities and are more widely used, there will be a great demand for highly skilled workers, such as data scientists. • Workforce Solutions Review • July-September 2017



Norbert Weiner, The Human Use of Human Beings, Houghton Mifflin Harcourt (US), Eyre & Spottiswoode (UK), 1950.



Frank Levy and Richard Murnane, The New Division of Labor: How Computers Are Creating the Next Job Market, Princeton University Press, 2004.


the brain, for decades there has been a vision to model computers after its workings. People learn by forming neural connections that are reinforced when a desired result is reached, and diminished when the desired result is not attained. Programming techniques have evolved to mirror this situation: using feedback loops, software is able to teach itself by seeking a goal and reinforcing desired results. We learn from our experiences; the analog to experience in the digital world is data. And, since we have so much of it, machines are able to learn quickly. This is a departure from past programming techniques, whereby the computer was programmed to do specific tasks. Now, a computer is programmed to learn on its own based on the data available to it. This technique is called “machine learning.”

change. Whereas managing people is primarily about getting them to do things they are disinclined to do, then making sure that they do them; managing a mix of humans, people enhanced by intelligent automation, and autonomous systems will require a more strategic mindset as to how to fit the assets together to optimize their effectiveness. There will be more emphasis on what humans do well that machines don’t. There will be more focus on how to include the people who fit together well personality-wise as well as skill-wise, determined by analytics. Human workers will be more empowered to make decisions and take ownership of their work. Thus, advancing technology will result in a change to how we manage people, as well as machines.

Organizational Impacts

There will be much scrutiny as advanced technology encroaches on the workplace. Legal considerations have always lagged behind the technology. As John Sumser noted in “Artificial Intelligence: Ethics, Liability, Ownership and HR” in this issue, if a machine makes a recommendation for action, does the machine incur any liability if the likely result doesn’t transpire? Just as licensing and regulation is par for the course with other knowledge-based services, the AI field should be proactive and provide full transparency and audit trail of all data pertaining to each recommendation, as well as the thought process used in evaluating that data. We must take into account the social implications of interacting in a deeper way with advanced technology. Technology has historically netted the same or more jobs. There is great debate over whether it’s different this time. Previously, machines replaced physical labor or repetitive administrative tasks. Now machines can perform knowledge work, which will make large sectors of the economy redundant. Not everyone will make the transition through the disrupted economy. We need to consider the social impact of many people being unemployable. There has been talk of the government providing universal minimum income. Some have suggested that robots pay taxes, in some form. Whatever the best solutions are, we need to be proactive in planning for conditions that may cause great social upheaval.

We saw earlier that we need to include contingent workers and even vendors in our workflows. We will need to account for both human and non-human workers in our organizational structures (by non-human, I’m including the possibility of animals: the police, search and rescue teams, and healthcare facilities have employed dogs, which are, indeed, knowledge-workers). Non-human assets will encompass automated agents of varying sophistication. Flatter organizations, with fewer hierarchical structures, will be needed. In the past, line level managers would supervise individual contributors, and each subsequent level involves the management of managers. That won’t be needed in a highly automated workforce, as automated agents will require a different sort of supervision than the traditional approach. More responsibility will fall on the vendors that provide the bots. Fewer people, and certainly fewer levels, will be needed to manage a human/robotic workforce. In his book from 1950, The Human Use of Human Beings,1 Norbert Weiner coined the term Cybernetics, pertaining to systems that self-regulate due to feedback loops, and also the interaction between humans and machines. Machine-learning algorithms use feedback loops to learn. The impact of cybernetics will play an increased role, as machines become a more integral part of our lives and our work. As humans, both employees and contractors will play a smaller role, and that role will

July-September 2017 • Workforce Solutions Review •

Legal and Social Impacts

Learning and Development

An important aspect of managing a workforce is the learning and development initiative. This is key to ensuring productive and fulfilled workers. However, as we saw earlier, machine learning systems are not programmed to do specific tasks; they are taught how to achieve desired outcomes based on their experience, which takes the form of massive amounts of data. Thus, machines need learning and development, just as human workers do. But, machines learn in a different way than humans do, and at different rates. People learn from their experiences, which unfold over a lifetime. Machines learn from data, which can be fed in at tremendous rates. Therefore, machines can have a lifetime of experience in a short time frame. As we saw in “Trading Cancer for Data: Machine-Learning for Cancer Diagnosis” in this issue, machines can be trained to diagnose cancer, or recognize a stop sign by being shown many examples, which are labeled “Yes” or “No.” As we develop personal assistants, it will be important that they be trained to have pleasant dispositions – so important in their interactions with their human counterparts. It will be like teaching a super intelligent child that has absolutely no common sense. Try teaching common sense – it’s no so easy. The way we educate humans will change also. As machines handle more and more tasks currently associated with knowledge workers, it will be important that we emphasize skills which humans have, but are still a challenge to automate. MIT’s Frank Levy and Harvard’s Richard Murnane argue that the automation of business processes has heightened the value of two categories of human skills: “expert thinking – solving new problems for which there are no routine solutions, and complex communication – persuading, explaining, and, in other ways, conveying a particular interpretation of information.”2 Our education is geared toward convergent learning – whereby there is one right answer, and our thought process is focused on attaining that answer. We should emphasize divergent learning – where there are a number of possible solutions, and we can imagine multiple options. Creative thought will be more highly valued. Artists develop the skills to perceive their environment differently. Although the goal of art is to realize one’s aesthetic vision, the process of creating art can be applied to solving business

problems. For instance, there are similarities between the process of composing a piece of music and creating software. The worker and manager will switch roles. It makes more sense for the machine to be the manager, freeing the person to be creative. I’ve known many artists and managers, and to my anecdotal observation, there is very little commonality among the skillsets. Effective managers must learn to get in touch with their inner muses.

Parting Thoughts

We need to change our philosophy as to how we manage organizations, encompassing human workers, business partners and technology, in preparation for a rapidly changing business environment. Beginning early to accommodate the current trends and technology will prepare us for the dramatic shifts in management that will be required in the not-too-distant future. Looking further down the road, opinions differ as to if, or when, machines will achieve consciousness. Regardless, we can expect rapidly increasing sophistication in machines. There has been much discussion (and some consternation) about when the robots no longer need us. At the hands of our much more advanced creations, are we in danger as a species? Anyone who’s heard Hal’s iconic, “I’m sorry Dave, I can’t do that,” from 2001 – A Space Odyssey, knows to always ensure that we have the off switch: ultimate discretion in an emergency situation. The industry must widely accept a set of principles for governing long-term advanced technology. This article covered a relatively wide range of time, wherein the way we work will dramatically change. The nature of what it is to be human will change as well. Although technology advances at revolutionary rates, human nature progresses at glacial, evolutionary rates. Many of humanity’s problems won’t be solved by advanced technology. Technology is a tool, which can be used to make life better or to manipulate and oppress others. We’re heading toward either Utopia or Dystopia (or maybe even both at once) as we weigh the cost-benefit of relinquishing privacy for convenience. Managing advanced technology is a tremendous burden; it is, in effect, managing the future of humankind. It may be a while before technology is as advanced as described in this article…but it’s closer than you think!

About the Author

Roy Altman is manager of HR Analytics and Architecture at Memorial Sloan Kettering Cancer Center. Previously, he was founder/CEO of Peopleserv, a software/ services company. Altman is the architect of multiple commercial software products. He has published extensively and has presented at several regional and global industry and academic conferences relating to HR and business process management (BPM), most recently at DisruptHR in San Francisco. Altman is an instructor at NYU, serves on the IHRIM Workforce Solutions Review editorial committee, and is on the board of Professional Exchange of HR Solutions (PEHRS). He can be reached at • Workforce Solutions Review • July-September 2017



Al Adamsen, Talent Strategy Institute

The Workforce of the Future: HR’s Role in Managing the Amoeba Organizations are comprised of people and the assets they use. They are formed at the outset of a business and over time to get work done. Employees do some of the necessary work. Other work is done by contractors, consultants, outsource providers, and alliance partners. Still other work is done by hardware and software – machines – and, increasingly, smart machines that leverage robotics, sensors, predictive analytics, machine learning, artificial intelligence (AI), and a host of other innovative technologies and analytical techniques. In this way, over time, an organization changes form, ebbs and flows, much like an amoeba. It might stretch to use more employee-centric power during one period then shrink that capability to leverage third parties or technology during another period. It might then split, acquire, layoff, hire rapidly, outsource, automate…any number of things to change into an appropriate form to meet market and environmental conditions. But, is the form an organization takes truly the appropriate organizational form or is it merely the one that comes into existence at a certain point in time? The opportunity that’s available, and that is now a race to figure out, is how to systematically measure and manage work over time – work, regardless of who, or what, is doing it. The ones that do figure this out will find themselves with a distinct, hard-to-replicate competitive advantage. In the simplest terms, they’ll be able to get work done faster and more effectively than their competitors. This should provide ample motivation, yet the ever-expanding array of options on how work can get done is pushing many leaders to a deerin-the-headlights response: “It’s too complex.” “We just need to hire more good people.” “We just need to leverage best-in-breed technologies.” Needless to say, these responses either simply ignore the reality before them or serve as


July-September 2017 • Workforce Solutions Review •

wholly inadequate responses to it. This cannot stand. Leaders must now ask themselves questions like: • •

• •

How do all the components of work function together to get work done? What is our holistic understanding of how work will get done next month, next year, and over the next 2-plus years? What are the disruptions we’ll face and opportunities we can leverage? What can we gain by having an integrated, disciplined approach to designing, measuring, and managing work over time? How will we build such a capability? And, what will it take to sustain and enhance it?

Answers to these questions are often: “I have no idea,” “We just have to focus on the here and now,” or the classic, “We’re not there yet.” Much of the reasoning behind these responses is: “We haven’t needed such an approach to date, so why do we need one now?” This rationale is troubling. The world is changing at an ever-increasing pace, thus moving forward in the absence of a systematic process around work design and management will all but guarantee excessive waste, inefficiencies, poor productivity, slow innovation, and low employee engagement. Employees might end up being overworked, undercompensated, overcompensated, undertrained or mistrained in the wrong roles, or doing marginally value-added work. Contractor numbers might be too many or too few. Technologies could be inappropriate or underutilized. A host of sub-optimal outcomes not only could emerge – they likely will.

This has to change, but to what – and, who will take the lead?

Right now, in most organizations, there are four roles best suited to lead the “this-ishow-we’re-going-to-get-work-done” decisionmaking process. They are the chief human resources officer (CHRO), chief information officer (CIO), chief technology officer (CTO) and chief operations officer (COO). Ideally, a governance structure, supported by process and information, will align the agendas of these individuals. Can we add the CEO’s chief of staff to this list? Sure. A facilitator bringing these historically disparate functional leaders together on a recurring basis for a clear purpose would, no doubt, be advisable. But this is not always possible, thus two of these four are bestsuited to take the lead. They are the CHRO and CIO. Chief information officers, of course, are responsible for providing insight and ideas on how to improve internal processes, position product, accelerate innovation, etc. Then, they must leverage all of this information as a source of competitive advantage. It’s the people within the organization, those doing the work, who generate much of the data that supports the analyses and advanced analytics done within the CIO’s function. What data and information do they generate? Is it consciously created or is it collected after processes and technologies have already been implemented? Historically, in most organizations, the answer is after processes and technologies have already been implemented, thus people and processrelated data supporting internal analyses have often been poor, non-existent, or flat-out inappropriate. The good news is that this is changing through an integrated approach to employee experience design. A CHRO’s main contribution to an organization, arguably, is to ensure that the company has the right talent, in the right place, at the right time, at the right price, and for the right reasons. Implied in this statement and others like it is that the organization’s leaders know how much talent it needs relative to the work that needs to be done. Also implied is that there’s a clear definition of what “right” means. For our purposes here, we’ll say “right” means an organization can accurately craft a role, as well as accurately locate, assess, and place individuals with appropriate technical skills to do the work, and who possess appropriate

behaviors to positively contribute to the organization’s culture. Culture, of course, is everyone’s responsibility (and opportunity), yet consciously creating, maintaining, or enhancing it over time is often facilitated by the CHRO. Facilitating the design and delivery of organizational culture, then, is another of the CHRO’s main contributions. How is this done? The answer again: employee experience design.

Employee Experience Design

Employee experience design is simply consciously-crafting interactions that inspire certain thoughts and feelings over time. Contrary to how it might sound, it is not meant to be overly structured or contrived. Instead, it’s meant to help people feel seen, heard and empowered. It’s meant to inspire authentic expression and conversation so that people feel safe – emotionally, psychologically, and physically – so they can do their best work. In this way, employee experience design strives to humanize the corporate experience. It strives to understand what’s important to an individual (most often through the use of archetypes) and, in turn, to deliver on that need or desire over time. This, then, facilitates the stories that employees tell themselves and others about their employment experience. These stories, of course, become the culture; if the culture is positive, a host of benefits emerge: • • • • • • •

These stories, of course, become the culture; if the culture is positive, a host of benefits emerge.

Better engagement, Better innovation, Better productivity, Better retention, Better talent attraction, More speed, and Better financial results.

These positive outcomes stem from the deeprooted belief – the trust – that the organization and its leadership are not interested in employees as an expense to be minimized or “human capital assets” to be maximized, but as people with certain wants, needs, fears and ideas: people with families, people with student debt, people who want to grow, feel more secure, be healthier, and contribute in ways that align with their values and purpose. Understanding, documenting, and designing an employee experience requires that organizations think about the data and information that employees create, are exposed to, and that are absent. After all, data affects • Workforce Solutions Review • July-September 2017


behavior, both the consumption of it (looking at a rating or performance metric), as well as the creation of it (think automated speed limit signs on roadways). With all this in mind, a multi-disciplined team is required to design employee experiences. It’s not just HR. In fact, HR functions best as the facilitator, not the project or process “owner.” This is critical, as all leaders need to be involved in consciously-creating culture and customer connection. They can’t merely sponsor the effort. They need to own it. An overview of an employee experience design process is detailed below.

Whichever process an organization uses, whichever employee experience framework is employed, the unmistakable need is that the effort needs to be multi-disciplinary. Why? Because employee experience design isn’t enough.

Employee Experience Design is not Enough

The workforce of the future is not just going to involve employees. It’s going to involve contractors, consultants, third-party outsourcers, affiliate partners, etc. This is already a reality, of course, as was highlighted at the outset. Unfortunately, though, in most organizations, these multiple components are not consciously orchestrated. Procurement manages contingent labor and other talent augmentation relationships. Human Resources facilitates talent acquisition and development of employees on the payroll. InformationTechnology manages the technologies used by employees. Operations dominates the day-to-day interactions and prioritization of work. Finance manages the budgets allocated to these groups. Do these functional leaders get together on a periodic, recurring basis to identify and measure the work required, then craft the most efficient and effective means to do the work? Again, unfortunately, the answer is most often, “No.” Will this


July-September 2017 • Workforce Solutions Review •

change? It must. Employee experience design is not enough. While it’s clearly a domain requiring HR’s leadership, it’s hopefully also clear that it’s a unique gateway to a more ambitious, more impactful objective. Most, in the industry do not view it as such. They view employee experience design as a project, and a limited project at that: “We’ll improve the employee experience, and then move onto the next project, the next priority.” The hard truth is this, though: the employee experience will always evolve. Employees change. Environments change. Technologies change. Candidates change. Market realities change. Budgets change. The employee experience, thus, needs to be continually updated to accommodate these changes. To be more impactful, employee experience design cannot be absent of workload management and the tools used to do the work. It requires visibility into the nature of the work and amount of it, then clearly identifying the work that’ll be done by employees. This will then inform talent strategies: org design, workforce planning, talent acquisition, training, internal mobility, workforce analytics, etc. It will also inform IT, Operations, and other strategies. In the end, employee experience management will evolve into, or emerge as a sub-component of, overall work design and execution. It needs to.

The Design of Work

At the time of this writing, Amazon just agreed to purchase Whole Foods Markets. The immediate speculation is now whether or not Amazon will broadly deploy the same technology it’s using in its experimental Amazon Go store in Seattle. If it does, whether next year or five years from now, the move will drastically affect the number of employees per store and per revenue, as well as drastically affect the employee experience. In fact, speculation around this is already affecting the experience of Whole Foods’ employees. Some feel hopeful and excited. Others are concerned and cautionary. Whatever feelings emerge, leaders at both companies would be wellserved to consciously manage communications, related expectations, and how imminent changes will likely affect everything from employment status to employees’ day-to-day activities, e.g., will Whole Foods’ cashiers have a job? If such an approach isn’t taken, then the inevitable uncertainty will almost assuredly elevate anxiety, compromise productivity,

and adversely affect a culture that to date has been a model of success (20 straight years on Fortune’s 100 Best Companies to Work for List). Similarly, and this might happen even earlier, with the introduction of driverless trucks, trucking companies are now facing a likely massive disruption.

Even the medical industry has disruption looming. Historically, doctors collected data from a patient, made a diagnosis, and then prescribed a treatment plan. Very soon, it may be the case that doctors have very little role in data collection (heart rate, blood measures, etc.) and diagnosis (what the issue or potential issue is). Even the treatment plan may be largely designed, measured, and monitored by hardware and software, tools that both feed and leverage artificial intelligence. In this example and the ones highlighted before, what’s the role HR should play? Will HR even have a role? These questions need to be answered, as these examples are not theoretical. They’re either happening or imminent, at least to some degree. As such, the way work is designed and executed needs to be more systematically thought-out and managed. A way to do this is depicted in the The Work Framework graphic shown here. It suggests that employees are those providing a high competitive advantage and, thus, should be secured over the long term (“long term” being defined by each organization to reflect its unique reality and mission). Consultants are high competitive advantage, yet serve as a short-term resources. Contractors and contingent labor are relatively low competitive advantage and also short-term. Finally, outsource providers and automation are relatively low competitive advantage, yet deliver a value proposition of the longer term. The Work Framework identifies the key elements of the organizational amoeba. It helps leaders think through how an organization can

best ebb and flow over time: what to contract, what to expand, etc. The lines and arrows depict the macro trends in each dimension. In the end, leaders can form a work plan and measure, About the Author Al Adamsen is the founder and monitor, and adjust over time. executive director of the Talent Strategy Institute. He’s one of Summary and Signals the few who’s led a workforce planning and analytics function The workforce of the future cannot be within a Fortune 500 comthought of independently from those other pany (Gap Inc.), served as a people, those other entities, and those other leader with an analytics vendor things (robots, AI, etc.) doing work on behalf of (Infohrm, now SAP/SuccessFacan organization. All of them affect the employee tors), and also as a consultant/ experience, the organization’s culture, its brand, advisor (with EY, Kenexa, and now TSI). Over his 20-plus year and its overall efficiency, effectiveness, and career he has served organizasuccess. If HR is serving as the facilitator of tions such as Disney, Starbucks, work design and execution, is that taking it out Chevron, T-Mobile, Boeing, of its “lane?” Not at all! It is expanding its scope SanDisk, Palo Alto Networks, and influence, yet it’s making the function more Mayo Clinic, Stanford University, relevant, powerful, and impactful. Plus, no other among many others. Adamsen co-chairs the People Analytfunction has yet filled the demand for this role ics Innovation Lab and People and, frankly, no other function is as well-suited Analytics Accelerator. He’s also a to do so. Thus, as more organizations choose to founding member of the Global use employee experience design techniques, the People Analytics Network. His opportunity arises to ask and answer some very educational background is in economics and individual, team, basic, yet increasingly insightful questions: and organizational behavior. As important, he is also a long1. What’s the work that needs to be done? standing coach to youth and 2. What will it take to get the work done high school coaches, parents of (capability)? young athletes, as well as young athletes themselves. He lives 3. How much time will it take to do the with his family in Santa Cruz, work (capacity)? California, and all are avid beach 4. Who or what will do the work (the work volleyball players. He can be plan)? reached at 5. Where will the work be done (workplace and location strategy)? As a final example, if the answer to this last question is, “On the organization’s property,” then is it advisable to have an onboarding process for employees, contractors, and consultants alike? All interact, thus contribute to the culture. It would be beneficial, but our functional, siloed thinking has anchored what’s possible to an old mindset. This mindset has to change. It has to shift to a more integrated, systematic way of thinking, one that looks at a diverse array of how work can get done. The amoeba is changing ever more rapidly, and like an amoeba, an organization’s primary mission is to survive; and to survive, it must adapt. Will HR adapt? Will leaders adapt? Will you adapt? The choice is ours. If HR doesn’t lead the way, it simply won’t happen, putting the organization, its stakeholders, and especially its employees at risk. Let’s take charge. Let’s lead. • Workforce Solutions Review • July-September 2017



John Sumser, HR Examiner

Artificial Intelligence: Ethics, Liability, Ownership and HR Enterprise software product liability. Those four little words make some people very uncomfortable. For the entire life of the software industry, no one ever really thought about product liability. The issues were all covered in the unreadable terms and conditions we were required to click through. Today, things are changing. The first era of software stretched from ballistic tables and payroll to the latest in HR forms completion. Over the course of 80 years, software recorded, collected, calculated, and reported data. Garbage in, garbage out (GIGO) was the primary principle. There could be no liability, because machines simply reported what they were given. In this second era, things are different. Some of today’s tools (and all of tomorrow’s) do much more than record and report. They suggest, recommend, decide, evaluate, prescribe, filter, analyze, monitor, and learn. Era 1 tools could not hurt people. Era 2 tools can. In a world where machines extract truth and insight, they (at least) share responsibility for the decisions they make. It may be that they have an exclusive right to the liability. One industry leading CEO says, “We call it machine learning when we talk about it internally. We call it artificial intelligence when we speak to the market.” For the purposes of this article, I’ll use the terms machine learning, artificial intelligence (AI), and big data somewhat interchangeably. I am referring to our computers’ emerging ability to change their output based on insights derived from new data. Human Resources enterprise software tools will be at the forefront of the implementation of new product liability concerns. Increasingly, HR software recommends and directs the behavior of managers and employees. If the guidance or insight is damaging or wrong, software


July-September 2017 • Workforce Solutions Review •

vendors will be unable to wriggle out from the consequences. Currently, the vast majority of recommendations provided by intelligent HR software are “self-correcting.” They learn from their mistakes and correct the underlying worldview. It is common to hear them described as tools that “get better with usage.” Another way of saying that is that their error rate improves over time. One part of the liability issue is the question “who bears responsibility for the error rate?” There is a more difficult dimension. All cultures, organizational or otherwise, are defined by their biases. The essence of culture is its unique worldview. Decisions and behaviors that support and expand the worldview are rewarded. Things that undermine or contradict the worldview receive negative feedback. Since algorithmic decision-making adjusts to the things that make a culture different, they tend to amplify the biases of the culture. Most of these machine learning tools are black boxes. The only way to see the bias is by examining the output. In other words, these new tools may create liability before it can be discovered and managed. Intelligent machines (and wonderful theories about their near and long-term potential) are at the heart of today’s pop culture. Everyone “knows” that self-driving cars are just around the corner, that robots are going to take your job, and that pretty soon you’ll be talking to an intelligent assistant who knows where you put your car keys and can order more laundry detergent. Yet, these programs and machines are not people who are generally governed by standards of reasonable care. They are property, which is governed by laws of strict product liability, warranty, and whether the design is fit for the intended use. The difference is much like an

owner’s liability when their dog bites someone. The dog is a separate actor, the owner did not intend for the dog to harm anyone, but the owner is strictly liable because the dog is the owner’s property. We are seeing companies develop and sell technology based on machine learning processes that the human designers do not fully understand or control. These machines are giving recommendations and suggestions based on probabilities to employees, many of whom are ill-equipped to understand and effectively use the information. Often, we don’t know whether the information will be useful until we develop and test it over time. But, it’s not just data. It’s evidence. And the laws that will apply are not the ones that organizations have traditionally been operating under. There is a chain of potential liability that runs from the developer vendors through the sales chain to the organizations using the software and machines. Not surprisingly, HR has been slow to adopt AI and machine learning technologies. The Sierra-Cedar HR Technology Industry survey1 suggests that fewer than seven percent of the companies they surveyed are using or considering using machine learning technologies in HR. Given that we are in the earliest of early adopter stages, it’s a solid time to think about the ethics involved in using machine learning systems to manage, supervise, assess, train, deploy, or categorize human beings. Ethics involves questions of right and wrong. Large institutions are usually concerned with optimization, efficiency, and innovation. They seek ways to maximize returns and minimize costs. They think of their employees as resources first and people second. Corporations have traditionally had some challenges with the very idea of ethics. Here are the kinds of ethics questions that are going to occupy the conversation about ethics in HR technology. The questions are standard. The answers may vary from place to place. • Who owns employee data? Who can sell or manipulate it? How much is an employee entitled to see? If something is wrong, what recourse does the employee have? If it is embedded in some machine learning scheme, what are the ownership variables?

If the system learns about itself through data about an employee, does the employee own the learning? Can he or she take it with them when they leave? Are they entitled to royalties if the data is sold for benchmarking or other purposes?

Is it okay to use data that was built without a control group? How do we measure effectiveness in the example of experimental control? When machines improve their error rate while implemented in an organization, are there extra employee protections required?

What is the line between manipulation and motivation? If chatbots increase emotional ties, is it okay to use them to increase engagement scores? What are the likely regulatory responses to overly manipulative work environments? Don’t people usually follow orders from authority figures? Doesn’t this apply to machines? Is there a limit to self-congratulatory positive feedback?

Are statistics more reliable than human decision-making? Where’s the proof? Is empathy a necessary part of decision-making? Kindness? Humans are demonstrably more effective at handling novel and/or erratic inputs to decision-making. Does that matter? How do you factor “unmeasurables” into the decision-making process?

How do you disagree with a machine’s decisions? Can you afford to be the person who is carping about decision quality? How do you get into a position to see bias? (It won’t show up in individual recommendations, it’s systemic.) Do we get stupid in the face of a machine recommendation? Are people predisposed to follow the instructions of an authority figure? (Consider Google Maps and one’s ability to argue with its recommendations.)

Who has the liability for machine recommendations? Who pays for damage caused by the machine? How do you handle mistakes? How do you monitor the quality of the algorithm’s performance? Is it ethical to use tools • Workforce Solutions Review • July-September 2017


that are known to be imperfect on employees? Are there implicit human experiences that are interfered with when machines are the arbiters of personnel decisions? •


Sierra Cedar Survey http:// research/annual-survey/


How do you limit the data’s ability to influence the company? How do you turn it off and replace it? How do you know when you have too much influence from a single source? Are there tools that allow you to see the risk in machine-led decisions?

In a nutshell, the ethics questions we will be grappling with are rooted in the fact that we simply don’t understand in a sophisticated way: • How human beings work, • How organizations work, and • How the human-machine interface will change these things. We are going to learn more about each issue in an accelerated way. The machines are coming to make your employment decisions, and we will learn from their mistakes and successes. About the Author You can expect to discover new ways of thinking John Sumser is the principal about employee safety as the risks at work shift analyst and editor-in-chief from physical to mental and emotional. of the HRExaminer. He is If you are considering the utilization of an independent analyst intelligent machines in your HR/Operations covering the entirety of the HR technology ecosystem processes, here are some questions you might from payroll and benefits to consider. recruiting. He has a particular 1. What are your views on product focus on ethics and practices liability? Be sure to have a long in predictive analytics. Sumser conversation about how the tool works routinely advises Human and how the vendor is monitoring the Resources and Recruiting departments and talent impact of machine learning curves. management vendor teams You’ll learn a lot by raising the topic with product analysis, market of product liability. Most vendors segmentation, positioning, still imagine that we are in the first strategy and branding guidance. generation of software where liability He’s been published and quoted in every imaginable outlet from is not really a possibility. The key The New York Times to HR question here is: “What if your tools industry trade magazines. He recommendations cause damage to is currently experimenting with people or our business?” the use of Facebook as a forum for industry dialog. Follow him 2. How do we make changes to there. You can also reach him at historical data? Most machine learning systems are “black boxes.” If you ask the designers how they work, they can only explain about 80 percent. That means that you are likely to want to modify the results that the machine


July-September 2017 • Workforce Solutions Review •

produces. It is likely that the answer to this question is: “You can’t.” Having the conversation is what’s important. It will give you a window to your real risks. 3. What happens when we turn the “it” off? How much notice will we receive if you turn it off? Imagine that you are using a tool that does the job of several employees (sourcers who review résumés, for example). If the tool fails in a way that requires a shutdown, what sort of advance warning do you get. Since most providers are in experimental stages, the answer to this question also matters if the project ceases to operate. In a very real way, these are digital employees, and it is best to have a replacement plan. 4. Do we own what the machine learned from us? How do we take that data with us? Part of the way that these systems operate is that they learn in both the aggregate and individual case. Most vendors guarantee that your data is “anonymized.” You still may not wish to have your operating practices be a part of some larger benchmarking process after you change suppliers. Being very clear about whether the system will retain evidence of your participation after you go is of strategic importance. 5. What are the startup costs, resources, and supervision? We know precious little about the behavior of intelligent machines. There is good reason to expect that their impact on your resource consumption is greater than anyone thinks today. Like any employee, they require training, supervision, and discipline. Make sure you have a very clear picture of the total cost of ownership of any leaning machine you enable. The age of human-machine integration is in its infancy. It is inevitable. In the transition, it is important that we move forward carefully with a clear picture of the risks and ethical issues. This note is a starting point. (Author’s Note: This article is the direct outgrowth of a presentation that I did with Stacey Harris from Sierra Cedar.)


Janet Mertens, IBM Institute for Business Value

IBM Study: More than Half of CHROs See Cognitive Computing as a Disruptive Force in the Next Three Years It’s clear that today’s organizations face increasingly complex workforce challenges. The virtualization of the workplace, a growing demand for novel skillsets, and a continuing stream of new technology and data are converging with a workforce that has heightened expectations for a compelling employee experience. The HR function has a key role in addressing these challenges by reducing complexity to create dynamic, engaging experiences that empower and enable talent. Several new technologies, including cloud, mobile, and the Internet of Things (the networking of physical devices such as sensors, wearables and other electronics), are helping to guide this ongoing HR transformation. Cognitive capabilities can further advance the evolution of HR by expanding human expertise and improving employee experience.

What can a cognitive system do? •

Understand: Cognitive systems can receive and process unstructured information in ways similar to those of humans. They understand language patterns and sensory inputs, including text, pictures and auditory cues. As such, they can interact with humans naturally. For example, a cognitive system can quickly examine thousands of hours of HR service center recordings to identify key words and patterns based on frequency, tone, and sentiment. Reason: Cognitive systems grasp underlying concepts, form hypotheses, and infer and extract ideas. They rapidly synthesize information to produce relevant and meaningful responses. Consider the case of a manager who

is looking to fill an internal role: A cognitive system could look at various data sources, including a candidate’s professional experience and previous performance, and then further analyze the candidate against the characteristics of other successful job holders to determine if he or she would be a strong fit for the organization. •

Learn: Cognitive systems learn and improve through every data point, interaction and outcome, building a deep and broad knowledge base that is always up-to-date. In the HR world, with a constant stream of changing policies and new regulations, this capability becomes critical. Rather than addressing a static set of rules, cognitive systems read, tag, and organize HR content from a variety of sources, allowing employees access to the most accurate information at any given time.

To better understand the impact of cognitive solutions on the HR function, the IBM Institute for Business Value, in collaboration with Oxford Economics, surveyed senior HR executives, CEOs, and employees across a range of industries and geographies. As part of a larger IBM global survey of more than 6,000 executives, we asked nearly 400 CHROs about their current views on cognitive computing; we also sought input from employees regarding their willingness to receive guidance from cognitive solutions. Our study reveals that the market for cognitive solutions in HR is set to increase dramatically over the next three years: Sixty-six percent of CEOs believe cognitive computing can drive significant value in HR, and almost • Workforce Solutions Review • July-September 2017


40 percent expect their HR function to adopt cognitive solutions during that time. Business leaders understand that this is a critical differentiator in the ongoing war for talent. Chief human resources officers are aligned with their CEOs; more than half recognize that cognitive will be a disruptive force in their industry. In fact, CHROs from our survey identify four key HR challenges that cognitive solutions can address (see Figure 1). Each of these challenges represents an opportunity to impact the bottom line – either through direct measures such as labor cost management and HR process optimization or through indirect means such as time-to-productivity and employee engagement. Human Resources executives from outperforming organizations appear to be even more Figure 1. Key HR challenges best addressed by cognitive computing.

aware of cognitive computing’s potential value in numerous HR disciplines (see Figure 2). Companies that report higher performance see strong potential for cognitive computing to address new and diverse challenges across a wide variety of areas. For example, more than twice as many outperforming organizations recognize the value of cognitive computing in talent acquisition. Our findings suggest that business and HR leaders recognize that cognitive computing will play a critical role in the future of human resources. However, many emerging technologies fail to reach their full potential because the workforce is either unable or unwilling to successfully embrace them. Given the potential transformative quality of cognitive computing, it’s important to assess the willingness of employees to interact with cognitive solutions in their daily work activities.

Figure 2. Belief that cognitive computing can add value in specific areas of HR (outperforming versus underperforming organizations).

The following scenarios were provided to respondents to gauge readiness for cognitive computing in HR. Scenario 1 – Benefits optimization: Buying extra vacation. As an employee, you have a chance to buy extra vacation, but are informed that it is unlikely to be approved as many others have already booked vacation. Would you apply for the leave, based on the advice provided?

Scenario 2 – Onboarding: New hire support After one week as a new hire, you feel the need for more support to help learn your responsibilities. You are advised that a new hire webpage contains a lot of useful information. Would you visit the website?

Scenario 3 – Personal coaching: Voice analyzer You have an important meeting scheduled with your manager immediately following a client call. After the call, you receive feedback that you seem anxious and should take a break before the meeting. Would you heed the advice and take a break?

Scenario 4 – Training: Team training The business wants to take a more systematic approach to employee training. As a team manager, you are provided a list of training opportunities for team members. Would you share the provided information with your team?

Scenario 5 – Selection: Candidate selection As a hiring manager in a large company, you discover the company’s recruitment approach is falling short because it interviews too few candidates. Would you start increasing your candidate list in the future?


July-September 2017 • Workforce Solutions Review •

How a Cognitive-enabled Approach Aids Human Resource Professionals

For cognitive HR to take hold, employees need to be comfortable taking advice from cognitive applications. To determine the willingness of individuals to engage with and derive insights from cognitive systems, we examined the responses of more than 8,600 employees to a series of typical HR-related scenarios (see sidebar on page 28). Each scenario described either a cognitive-enabled approach to support a decision – a mobile cognitive chatbot, for example – or a traditional HR source of information, such as an e-mail exchange with a manager. Responses to each scenario include a “desired decision” from an HR perspective. For each scenario, we compared responses across several dimensions including: • Do employees make the same decisions when advised by cognitive systems versus traditional HR professionals? • Do employees feel as well-informed by cognitive solutions as traditional HR strategies? • To what extent do workers trust information from cognitive systems versus traditional HR processes? Our employee research revealed the following: Behavioral intentions – Respondents indicated they would make similar decisions regardless of whether they received advice from traditional sources or cognitive solutions. This suggests employees are able to glean appropriate information from cognitive systems. The biggest difference we observed in this trend was in the voice analyzer scenario. When respondents received traditional human advice, 60 percent reported an intention to make the desired decision, compared to 56 percent that reported an intention to make the desired decision when the advice was from the cognitive solution. Information adequacy – An important question relates to whether there is any informational advantage offered by cognitive solutions. When asked if they had sufficient information, respondents who received information from cognitive systems tended to answer “yes” more frequently (68 percent on average across the scenarios) than respondents given the traditional advice (64 percent on average across the scenarios). This

difference was especially pronounced for more complex decisions, such as whether to buy extra vacation. In this scenario, 58 percent of respondents reported having sufficient information from cognitive, compared to 50 percent of employees in the traditional scenario. Trust – We also examined the level to which respondent trusted the information they received. Two scenarios showed noteworthy results; the highly complex vacation scenario and the personal voice analyzer scenario. In the complex vacation scenario, people trusted cognitive more than traditional (58 percent versus 54 percent). In contrast, in the personal voice analyzer scenario, people trusted traditional more than cognitive advice (68 percent versus 57 percent). It is evident that employees have similar levels of trust regarding information received from cognitive applications and information from humans, especially where decisions are complex, but less so when decisions are personal. Intent to reuse – Our research shows that participants who received traditional support expressed a slightly stronger likelihood to seek similar advice in the future, compared to participants who received cognitive support. Nevertheless, the proportion of respondents indicating they would re-use cognitive solutions was still relatively high across all scenarios, ranging from 47 percent to 73 percent. This suggests there may be a short learning curve as employees build familiarity with cognitive systems and learn to make full use of their features.

The Cognitive Sweet Spot

Our research clearly demonstrates that organizations are primed for cognitive HR, and employees are ready to embrace it in many daily activities. However, there are clear indicators for success, making it important to recognize the “sweet spot” where cognitive solutions will have the most significant impact (see Figure 3). The cognitive HR sweet spot occurs when: • Decisions are information-rich and highly complex – requiring a wide variety of inputs from different data sources. •

Interactions by users are frequent and varied – where large volumes of requests must be interpreted in different ways.

High volumes of unstructured information are involved – such as free form text, graphics and images and auditory cues. • Workforce Solutions Review • July-September 2017


Figure 3. The Cognitive HR “Sweet Spot.”

The output is expected to be customized and personalized – to address the individual needs of a global and diverse workforce.

In the world of HR and employee experience, we identify three areas well suited to take advantage of the benefits cognitive capabilities offer: • Talent acquisition and onboarding – Cognitive solutions can tap into multiple data sources and reveal new insights to help companies develop richer candidate profiles, position themselves more effectively in the external labor market, and make better decisions about prospective employees. • Talent development – Cognitive insights can produce more personalized recommendations for learning and career management. • HR operations – Cognitive computing can enable more streamlined and accurate information by equipping and empowering HR advisors.

Beginning Your Cognitive HR Journey

Taking the first steps toward introducing cognitive capabilities to your organization need not be a daunting task. Here are recommendations for how to begin. Consider how cognitive will strengthen your HR transformation. Cognitive capabilities amplify existing investments across HR, including core HR platforms and other cloud-based applications. For example,


July-September 2017 • Workforce Solutions Review •

in talent acquisition, consider how a cognitive system might better predict new hire success by examining candidate data in combination with your organization’s internal performance metrics and competency frameworks. In talent development, look at how cognitive applications could fortify your learning management system to help guide learners toward up-to-date information aligned with the strategic priorities of your business. Finally, in HR operations, determine how employee self-service could benefit from a mobile chatbot that interacts naturally and on demand. Start simple, but start smart. Deciding where to begin with cognitive computing can seem formidable. Consider which cognitive capabilities are best suited for the problems you want to solve. Natural language processing capabilities are valuable in situations where repetitive interactions with the employee population are required, while the ability to interpret tone is a valuable asset for managers and leaders. Consider the cognitive “sweet spot,” which involves situations where decisions are information-rich, highly complex and frequently required by employees. Focus on opportunities that will benefit from novel insights, incredible experiences, enhanced expertise or intelligent processes. Build an interdisciplinary team to co-create solutions with a broad selection of your user population. Understand the possibilities of your data. Because cognitive systems excel at uncovering insights from sources that were once unsearchable, traditional text-based data can be augmented with sensory inputs such as natural speech and images. Unlocking the potential for differentiated insights requires understanding the data you possess, as well as data outside your firewall and data yet to come. When asked which data sources would be most important for use in cognitive solutions, HR executives from our survey identified external labor market sources (46 percentof respondents), internal HR data (46 percent) and employee competency models (45 percent) as the top three. Build trust and engage people. Understanding the implications of the integration of people and machines in the workplace must be an essential part of your cognitive journey. Over half of the HR leaders we surveyed recognize that a wide variety of HR roles, ranging from senior executives to the employee service center, will be impacted. Consider the far-reaching implications, including the antici-

pated need for reskilling and potential job redesign. Preparing people for new ways of working with technology is a foundational step; adapting processes, content and roles helps pave the way. Convincing employees to use the cognitive solutions is the next step. Build trust by focusing your initial forays on systems that augment and support your employees’ expertise. Enhance and expand strategically across HR. Cognitive systems are designed to learn and improve. Plan to assess the system’s progress and continuously apply feedback to enhance and deepen cognitive functionality. As you refine and grow, assess your progress and measure the specific value of each solution. Being deliberate in your progression across different areas of HR, including talent acquisition, talent development and HR operations, can help you realize value in many elements of the employee experience.


Are You Ready for Cognitive HR?

This analysis and discussion is taken from the IBM Institute for Business Value study on cognitive HR, Extending expertise: How cognitive computing is transforming HR and the employee experience. To learn more about how cognitive computing can transform HR in your organization, download the full IBM report at The study explores how key functions of HR can benefit from cognitive solutions, and highlights companies that are already leveraging cognitive capabilities to strengthen the employee experience and improve HR service delivery.

Free Personalization 48 HR Express Service 1000+ Product Styles!

Awards. Trophies. Gifts.


About the Author

Janet Mertens is global HR research leader for IBM’s Institute for Business Value. With more than 15 years of experience in organizational effectiveness, enterprise learning and human resources, she provides guidance and leadership to organizations undertaking major transformations, and champions the development of exceptional stakeholder journeys. Her most recent publications have focused on the impacts of technology on the workforce, and the importance of the employee experience in business performance. Ms. Mertens can be reached at • Workforce Solutions Review • July-September 2017


2017 Mid-Year Source Buyers Guide The 2017 Mid-Year Source Buyers Guide will serve as a valuable reference tool. For your convenience, the guide has two sections: a Categorical Listing and an Alphabetical listing. In the Categorical Listing, companies are listed under the product and service categories of their choice. For information on a specific company and its products and/or service, please refer to the Alphabetical Company Listing. While a listing in this guide does not constitute an endorsement by IHRIM, it does indicate that these companies are interested in serving the needs of HRIS professionals. We hope this Buyer’s Guide will assist you in your 2017 purchasing decisions.

Product Categories


Optimum Solutions StarGarden Corporation

Compensation Management

Deferred Compensation DECUSOFT Executive Compensation DECUSOFT Incentive Compensation Enterprise Information Resources Inc. DECUSOFT


Paid Advertising

HR Service Delivery

Cloud Computing Enterprise Information Resources Inc. Optimum Solutions Self Service Optimum Solutions



Self Service

Employee Self-Service (ESS)/Manager SelfService (MSS) Telliris

Payroll Software

Optimum Solutions

Optimum Solutions Telliris

Performance Management

Enterprise Information Resources Inc.

July-September 2017 • Workforce Solutions Review •

Time and Attendence Systems

Alphabetical Company Listing* *Systems and applications referred to in this section are trademarked, registered, or in progress. These names should not be used generically.


18475 E. Valley Blvd. City of Industry, CA 91744 Michelle Smith 888-779-8803 888-669-0838 is a leading supplier / manufacturer of crystal awards and corporate gifts. We offer free engraving and no setup charges on all of our crystal awards and gift products. We have in house professional graphic designers, engravers and customer service specialists to serve our customers making ordering crystal awards and gifts easier than ever. At Factory direct prices and with huge inventory selection at our California warehouse, you can’t find any better prices and faster turnaround for the same premium quality of custom engraved corporate awards, sports trophy and personalized gifts. See our ad on page 31.


70 Hilltop Rd. Suite 1003 Ramsey, NJ 07446 Michele Weiss 201-258-3395 201-785-0774 You have an HCM software suite but you are managing compensation outside the system. Now what? You need COMPOSE, a specialized compensation management software solution that handles any level of variable compensation complexity, reduces your total cost of compensation administration and integrates with existing HR Solutions. Not so suite, but oh so right. See our ad on the Inside Front Cover.

Enterprise Information Resources Inc.

StarGarden Corporation

271 Waverley Oaks Rd. Suite 207 Waltham, MA 02452 Gin O’Leary 855-589-9451 617-924-4802 www. Enterprise Information Resources Inc. (EIR) Get the most from your talent management strategy with EIR expertise, proven technology and service offerings. EIR DataTools are advanced automation tools that turn your system into a major company asset providing the accurate, actionable data necessary for reaching a true competitive advantage. EIR is a member of the SAP PartnerEdge program. We are authorized to resell and are a certified implementation partner for SAP SuccessFactors solutions. See our ad on the Inside Back Cover.

Optimum Solutions

210 25th Avenue North, Suite 700 Nashville, TN 37203 Scott Henderson 615-329-2313 615-329-4448 Optimum Solutions provides Payroll, HR, and Time & Attendance software delivered on-premise or in the cloud (OptiCloud®). All applications are developed and supported internally, giving your company the individual attention it deserves while providing you with a complete, one database HRIS solution. Optimum Payroll clients currently process over 12 million paychecks annually.

300-3665 Kingsway Vancouver, BC V5R5W2 Marnie Larson 800-809-2880 Let StarGarden’s 30 years of experience help manage your most important resource. Get the right resource on task at the right time with StarGarden’s advanced HCM, Payroll, and Workflow functionality. Visit us at to learn more.


4 Armstrong Rd. Shelton, CT 06484 Sales Department 203-924-7000 Mobile Enable your Time & Attendance with Telliris. It’s integrated and ready to use with many packages (Ceridian, Focus, HBS, Identatronics, Infinisource, InfoTronics, Insperityy, Kaba Workforce Solutions, Kronos, Renova, ScheduleSoft, Sense Software, SumTotal Systems, Time Link, UniFocus, and Workforce Software). • Workforce Solutions Review • July-September 2017



Freddye Silverman, Silver Bullet Solutions

Bots ‘R Us “Hello Google, what time is the next showing of Guardians of the Galaxy? Siri, what Italian restaurants are near me? Alexa, please turn on the kitchen lights.” If I had asked you just six short years ago if you thought anyone could have an intelligent and verbal personal assistant on hand to answer questions, assist with transactions, and remind you about everything just like your mother does at any given moment, you might have thought it was nothing more than a gleam in an engineer’s eye. Let’s go back to the turn of this century (which, in tech years is probably that number squared). Back in the day, the technology you used at work was typically more advanced than what you used at home. Of course, PCs were ubiquitous, but remember how cool we thought Blackberries were? At my company, being issued a corporate blackberry (for email, not just the original pager) was truly the brass ring, and signified your membership in the upper echelons of management. Self-service apps at your workplace were also considered high-tech, and the words “consumerlike experience” weren’t yet in use. We were all still shopping at the mall…and still going to movie theaters. The pace of technological development continues to astound, and even those of us who work in the technology arena are often hard-pressed to keep up with the latest developments. (A confession…I’m a bit of a Luddite despite my years in tech. They had to pry my beloved Blackberry from my hands before I gave in and got an Android, which I thought was too big and much too complex for my needs. Now you can’t take my Galaxy away.) The work/home technology experience has significantly shifted over the past decade with business-to-customer and personal tech out-distancing the corporate tech arena by years. Many businesses are still waging the social network battle and attempting to prevent workers from using all of the technology that is available to them outside the workplace during work hours, clearly a losing battle. Well beyond the basics, the newest frontiers are the “emerging” technologies of artificial intelligence (AI), chatbots (intelligent assistants), augmented reality (AR) and drones. The quotes around “emerging” are intentional; while that term is broadly used, especially in HR, these technologies are, in fact, well past the cocoon stage. They are, and have been, fully developed


July-September 2017 • Workforce Solutions Review •

butterflies for quite some time – just not in the HR world. That fact is even obvious in one of the names – it’s no longer good enough to simulate reality in a “virtual” way. Now, it has to be augmented, ratcheted up a notch or two for a truly wow experience. So AR is already clearly in version 2.0 or more. “Emerging” is a relative term – the technologies are not new, but their broad application in HR is. Artificial intelligence is moving farther along the transformational path, certainly at home and in work operations, but still in its early stages in terms of HR. Cognitive computing is the simulation of human thought processes in a computerized model that involves self-learning systems. These systems use data mining, pattern recognition, and natural language processing to function like the human brain. Chatbots are the user-facing channels, intelligent assistants with high-quality voice simulation and conversational ability. Competitors in this market range from IBM’s Watson to human capital management providers and many small startups that are targeting specific areas of HR, including recruitment, service center Q&A and personalized learning. Just as Watson ingested and analyzed all of Wikipedia in order to play – and win – Jeopardy, Bots/AI could scan all employee insurance policies to answer questions during open enrollment. We are in an era of post-mobile development where no device is necessary and search is no longer bound by text; instead, transactions are conversational. Google is investing vast amounts in machine learning and the belief that Alexis-like assistants will be standard issue, but via an implanted chip in you, your house and/or your car. SRI, the company that invented Siri, has developed a platform called SenSay Analytics, which can detect and respond to emotions. It parses through words, tone, volume, pitch and other characteristics of the human voice. When it senses anger, it can apologize, will speak more quickly when it senses impatience and will forward you to a human agent when it senses that you are on the verge of a major blowup. This would be a logical tool for an employee service center and be another branch of Tier 0 (along with self-service) since no humans are involved. Per SRI, the product can sometimes miss picking up verbal cues and get it wrong – but so do people. Odds are, it could answer

most employee questions adequately and with great consistency. Talla, a Boston-area startup with approximately 2,000 installs to date, has introduced ServiceAssistant, which is designed for completing HR or a help desk’s daily tasks, such as explaining company policy, training new hires, surveying employees and collecting information. It operates inside messaging software such as Slack, Hipchat and Microsoft Teams. While it will replace lower level employees at the start, as the AI functionality is developed, it will replace less manual jobs. To date, the recruiting sector has seen more practical application of AI than any other HR area, such as: – A programmatic advertising company that focuses solely on the recruitment marketing space. You tell the program what candidates you seek, and it creates targeted campaigns across apps like Google, Twitter and Indeed. MosaicTrack (https://www. – Similar to a résumé scanner but it goes further; it simulates how a hiring team reads a résumé, and makes a yes or no decision. It reads chosen résumés based on job descriptions, learns how to match résumés to certain jobs, and gives recruiters better recommendations. Mya ( – This is a recruiting assistant that works with applicant tracking systems and does all the jobs needed throughout the interview process – prescreening, answering FAQs from candidates, provides application tips, etc. SAP has done significant development in chatbots and is planning to launch a suite of them at the end of 2017 that all function inside of the Slack messaging tool and focus on their Concur expense management tool, Hana and SuccessFactors. The employee can converse with the Concur bot, for example, to build a travel itinerary. The bot will only respond with flights that are within the company’s approved travel policy costs. If none are available on the requested date it will suggest another departure date with open flights that are within the price guidelines. At that point, the employee can tell the bot to book the flight. The Hana bot will inform specific employees when relevant tasks are completed, e.g., Sales Order 123 was delivered to Company X on June 1. Finally, the employee can speak to the SuccessFactors bot and tell it to give spot awards or kudos for jobs well done and the bot will confirm when the kudos are delivered. Where

employee self-service was the starting point of moving transactional HR to an automated platform, chatbots will eventually push that to the next level, taking the place of more staff roles. A prime example of the transformation that chatbots and robots can deliver is in Japan. That country is leading the way in robotics for several reasons. Due to its aging population, low immigration and low birth rate, the labor market has been shrinking. The growing elderly group needs help and services, which must be provided with less people. The cellphone company Softbank developed a robot called Pepper as a home companion. It is emotionally aware and can discern tones of voice and facial expressions. The home version was so successful that they developed an enterprise version for businesses, and the first job it will replace is a receptionist or greeter. The Japanese hotel chain, Henn Na (which comes from the root word for strange), opened the first five star, fully robot-automated hotel in a Nagasaki theme park in 2015 and met with such success that it has since opened two more. Humanoid robots greet Japanese, Chinese and Korean guests at reception; a robot dinosaur greets English-speaking guests (apparently the Japanese believe that Disneylike animatronics are preferred by Western guests).

Robots engage in intelligent conversation, carry luggage to all rooms and clean the rooms.

There are limited room amenities; guests request them via tablet.

Guest room doors are opened by facial recognition technology; the hotel is keyless.

Instead of air-conditioning, a radiation panel detects body heat in rooms and adjusts the temperature.

Rather than have fixed rates, guests bid for rooms during peak season with a price cap on bidding.

Solar power and other energy savers are used to reduce costs.1

The Henn Na hotel group plans to build 1,000 more hotels around the world. Just imagine the seismic shift in hospitality jobs if more hotels were to adopt this approach! Clearly, the advent of chatbots and robots is encroaching on all professions, not just factory workers, as it originally did. Experts have • Workforce Solutions Review • July-September 2017


Endnotes 1

Daisuke Kikuchi, “‘Strange’ hotel, run by robots, opens near Tokyo; more to come,” The Japan Times, March 15 2017, http://www.japantimes. business/strange-hotelrun-by-robots-opens-neartokyo-more-to-come/#. WTM6MMa1vIU.


“A New Trend in HR? Drones to Headhunt Smart People!,”, October 25 2016, http://www.branding. news/2016/10/25/a-newtrend-in-hr-drones-toheadhunt-smart-people/.

estimated that 47 percent of all jobs are subject to disappearance in the next decade. It has already eliminated some software developer jobs, store greeters, any job providing information – much like an automated kiosk. The displacement will take place in unexpected jobs as well. A headline article about a baseball game was written by a robot using an algorithm versus one written by a reporter. When surveyed, readers’ preferences were divided 50/50, which came as a great surprise to those who believed that journalism was one of those fields where people are not replaceable.

Christopher Trout, “British Airports Now Beaming Holographic Security Agents,”, Jan 2 2011, https://www. british-airports-nowbeaming-holographicsecurity-agents-video/.


Virtual reality is being used in HR in specific administrative areas right now, rather than with the entire employee base. One example is Fidelity Lab’s application to help HR managers keep track of employees’ retirement plans and status. Research has shown that those managers paid more attention and responded better to a virtual reality version of their employees’ retirement plans About the Author and status, which used data visualization versus Freddye L. Silverman is founder of Silver Bullet a two-dimensional spreadsheet. The application Solutions and an independent works with the HTC Vive headset and depicts their HR technology consultant. workers divided into seven sections by tenure at Prior to this, she was vice the company. president, Eastern Region, at Those in green were judged to have successful Jeitosa Group International, retirement planning; those in red had possible a Workday partner. In her last practitioner role as VP problems. The managers grasped the problems of HR Technology Solutions and came up with plans to address the issues much at Cendant Corporation, quicker with the virtual reality app, compared to she was responsible for technology strategic planning having the data presented to them in spreadsheet format. and oversight of global HR systems and the U.S. payroll As I write this, there is a conference called system. She has been actively “Drone Focus” being held in Fargo, North Dakota involved in IHRIM since 1986 (apparently, the Silicon Valley of the drone world!). and was its president in 1997. While there are sessions on Drones in Journalism She has been an adjunct professor in HRMS, has made and Drones in Construction, there are none on Drones in HR. But, don’t assume the lack of a many presentations, and published articles on related conference session means that drones are not topics both in the U.S. and already being used in the HR arena. In the city abroad. Silverman taught of Brno in the Czech Republic, a tech hotbed of Spanish at the secondary Eastern Europe, there is significant competition and university level for for talented employees, and companies have to 10 years before making a career change into IT. She find creative ways to attract top personnel. Kiwi. can be reached at freddye. com is an online travel agency that operates with the mantra of “doing something stupid” as the best way to publicize their business. One of their latest recruiting solutions is a fleet of HR drones


July-September 2017 • Workforce Solutions Review •

designed to catch the attention of the many developers in the city as prospective candidates. The drones carry a banner proclaiming “Smart People Wanted” with the Kiwi website and an email address. Their head of HR, Katerina Gabova, stated that, “Recruiting the best in the industry is always a challenge…smart people need to work somewhere that inspires them. We wanted to dramatically show that…we foster an environment in which clever people will thrive.”2 No stats are available yet to judge the efficacy of this approach, but I would not call it stupid at all, despite their mantra. On the contrary, it’s a singularly clever approach to recruiting their target audience. Finally, if you’ve ever seen a hologram or gone to a holography exhibit, you were no doubt impressed at how real the holograms are. You reach for a piece of candy in front of you only to find there’s nothing but air there. And, don’t equate holograms with virtual reality – you don’t need to wear any special device in order to see holograms. As early as 2011, two London airports beamed in holographic images of agents to help prep people waiting in the security lines. Not only did they replace humans, delivering the same messages repeatedly, but they also did so with consistency as opposed to their human counterparts.3 We haven’t seen it yet, but it’s only a matter of nano-time before someone weds the hologram with artificial intelligence to beam out lifelike chatbots, a.k.a., HR representatives to the factory floor, to remote employees, to the employee cafeteria, or relaxation lounge. That’s the ultimate combo of automated technology and the personalized user experience that can attract, engage, and retain all those coveted talented workers that companies vie for. I’d say that would go in the win column. Human Resources is, and will be, even more impacted by all of these technologies, both in their own departmental work, as well as their roles as business partner to managers and employee advocates. One of the most critical factors in selection and deployment of any HR system now is the employee experience. It is expected to match or beat the consumer experience we have at home, both in ease and timeliness. Artificial intelligence, chatbots, virtual reality, drones, and holographs – all of these will contribute to an incomparable and highly personalized user experience. Take that, Amazon!

Executive Interview

An IHRIM WSR interview with Catharina Lavers Mallet, COO and head of product development for Talla IHRIM WSR: Thanks for taking the time with us today, Catharina. To start, how would you best describe Talla for someone in the world of human resources? CAT: Talla is an intelligent chatbot supported by machine learning and artificial intelligence (AI) technology that acts as an extension of your HR or IT team. Talla takes care of all of the repetitive communications that a lot of HR people find themselves doing on a daily basis, answering the same questions and sending the same memos and reminders around benefits, policies and procedures. Talla learns from what you do as a service provider, and over time takes on some of that repetitive work. IHRIM WSR: How does Talla use artificial intelligence to help address these different types of inquiries? CAT: Talla learns from what people do to answer a question or a request, so the next time Talla is faced with that request, it knows what to do. Over time, Talla gets better and better at knowing what the right information is to give employees, given the context of that individual person. IHRIM WSR: Can you provide some background on AI and then give an example? CAT: Until recently, the use of natural language for AI was very rule-based, so you had to configure very specific steps, and if you varied from those steps, the whole thing fell apart. Machine learning approaches that have been developed in just the past few years are a lot more flexible and can handle all the variations of language more easily. Now, you can ask the same question half a dozen different ways, and given the advances in AI, it’s a lot easier for the system to understand that those questions are equivalent. For example, you might have an employee who asks what the company’s commuter benefits are. In the past, this employee would ask the HR manager, or maybe poke around some internal help documents and hope that information was still accurate. Now, they can use the company’s current chat system and ask Talla, “What are our commuter benefits?” Talla replies immediately with that information, and can even walk you through a form to sign up for it.

This improves the user experience, allowing the employee to focus on the work that is most important. IHRIM WSR: How has the rise of messaging platforms, such as Slack, influenced the development of Talla? CAT: There were two major shifts that are really influencing the development of Talla. The first one involves significant advances in the development of Natural Language Processing (NLP) and Natural Language Understanding (NLU) in the last few years. The second is a shift in the way that people communicate, with more of a focus on chat and conversational exchanges. People are used to instant messaging and chatting about all contexts of their life, and we’re seeing that much more in the workspace as well. And so, we use chat and conversation as a very strong driver, but we also support email, portals, and all of the ways that people currently work. IHRIM WSR: How long does it usually take from the time that a client says to you, “This is something that we’d like to do,” to the time that you have actually one of these assistants up and running? CAT: It’s really, really quick. It requires a few clicks to install, and then you announce to the team that you have a new way of routing questions to HR and IT. And, when you deploy, you truly see value from day one, and we see that within about 60 days, there’s noticeable relief for your administrators, your service providers, in the basic questions that they’re answering, which really frees them up to work on higher-value projects. IHRIM WSR: How does Talla know what to point to in terms of an employee, not in terms of an organization’s knowledge base? CAT: Talla learns from what it sees you do. If you are the HR service provider, the first time somebody asks a question, you go find that answer in whatever systems you have, such as a Wiki or Google Drive. Then, you basically point that question to that answer, and send that answer right back to chat. Talla learns where the answer to that question is located. So, a lot of the value is in understanding what the question is and saying, “Okay, I know where that answer is.” We are in the process of building out • Workforce Solutions Review • July-September 2017


deeper integrations into common HR and IT systems, so that we can also draw on contextual information about that employee. IHRIM WSR: What are some of the challenges of “teaching Talla” in terms of building that knowledge base within it?

About the Author Catharina Lavers Mallet is the COO at Talla, an early stage conversational AI company that is focused on improving productivity and communications for knowledge workers. She is a strategic and operational leader whose passion is helping young, product-led, data-driven Software-as-aService (SaaS) companies get where they want to go. Previously, she served as the London Studio general manager at King Digital Entertainment (developer of Candy Crush, and now part of Activision Blizzard), and held leadership roles at Playfish (acquired by Electronic Arts) and Algorithmics (acquired by Fitch Ratings), among others. She has an MBA from MIT Sloan and a B.A., cum laude, from Harvard University, and now lives in the Boston area with her husband and two young children. For more information about Talla, visit

CAT: I think from an adoption perspective, the challenge is that the team using Talla has to be committed to routing support through Talla. Talla only knows what it sees, and if things are being done offline and people are still doing what we call the “drive-by,” it’s difficult. The learning process is going to be slowed down a bit. But, what we see is that users are actually really receptive to using Talla, because it doesn’t require them to go anywhere else for information; and they get an answer faster than they would otherwise. IHRIM WSR: Are there any surprising employee reactions that you have seen during implementation? CAT: The thing that has struck me the most is how quickly people get it; they say, “Ah, okay, that’s basically what I’m doing already, but now I’m just sending the question to a bot instead of an individual, and then getting an answer faster.” IHRIM WSR: It sounds like, as you said before, it’s the commitment to routing the answers through Talla that is a critical part of the learning process. CAT: Right. IHRIM WSR: So what are some of the other lessons learned that you’ve gathered from some of these early deployments? CAT: We’ve learned a lot from deploying different chat-based products over the last year and a half. We’ve had over 2,000 companies use them, and so we understand well what works in a conversational interface and what doesn’t. Some things are best done in chat and some things just aren’t. For things like scheduling, we’ve learned that it’s faster to click a few buttons than it is to type out what you want to do, and who and where you want to meet. We found that Q&A and support issues are really great use cases for chat, and that’s really part of the reason why we’ve focused on this area. IHRIM WSR: What are some of the other areas outside of HR that you’ve been working in? CAT: Currently, we’re looking at HR, IT and facilities – generally internal service providers within the context of a company. We’re also being used to support individual teams. You might have a business team that wants to be able to automate some of their FAQs if they have a lot of people joining, for example.


July-September 2017 • Workforce Solutions Review •

IHRIM WSR: How do you think the world of chat bots is going to change HR? CAT: Human Resources can get really bogged down by all of these tier-one level issues. Using chatbots, automation, and artificial intelligence to free them up, lets HR focus on much more important, high-value initiatives, such as building employee engagement, talent development, etc. I think it also allows for a more holistic employee experience, letting people get help in the way that’s best suited to their working style. There are just a lot more options for the end user, whether they want to talk to an individual, chat, or email. I also think that every employee has 20 percent of their job that they wish they could automate away, and Talla and chatbots really help people do that. Usually, when you look to improve the employee experience that means a higher cost in some way, but with machine learning tools like Talla, you get a better experience at a lower price from a bot that can work 24/7. At the same time, more and more companies care about delivering a good employee experience. Using assistants like Talla helps people get access to information more quickly, and get back to doing their jobs. IHRIM WSR: Could you provide some information on the history of Talla? CAT: Talla’s been around for about a year and a half. From the beginning, we felt strongly that there was an opportunity to look at the intersection of artificial intelligence and chatbots within the context of an enterprise or within a company. We’re really focused on the knowledge that is held within a company and how we can continue to facilitate the exchange of information and communication. We started doing some work around recruiting, but we found that it’s really in the area of a help desk, or support desk, that we’re getting a lot of traction, and a lot of interest. People get it right away and see the value. IHRIM WSR: Is there a sweet spot in terms of company size where this starts to play a significant role? CAT: The really fascinating part of our customer development and product development cycle is that we see interest from companies with a 100 employees that like the idea of Talla as their first lightweight service desk or ticketing system. Then, we also see it as a solution for companies with 10,000-plus employees that need to relieve some of the demand on their tier-one support staff. We are exploring how best to address the needs of these different segments, so that they get the most value.

The Back Story Katherine Jones, Ph.D., Mercer

May the Bots be with you: RPA for HR Spoiler alert: People are fallible. Here are some interesting factoids: Business people typically make errors at work at a rate of 10 to 30 errors per 100 opportunities. The best performance possible in well-managed workplaces is an error rate of 5 to 10 in every 100 opportunities. The researcher concluded with two points: 1. Even with highly experienced and able people doing the work, there are excessive rates of failure. 2. Letting people work from experience and knowledge always creates unwanted random variation that too often produces wrong outcomes.1 That leaves room for a huge amount of error. What are the options? In the medical arena today, computer-based cancer diagnosis using an algorithm is 81 percent more accurate compared to the consistency of the diagnoses of pathologists – which is only 48 percent.2 In the practice of diabetic retinopathy, an algorithm created to analyze retinal images matched or exceeded experts in diagnosing 12,000 images.3 The potential for accuracy, consistency, and compliance increases with smart machines and intelligent software – and unlike humans, computers never get tired or careless. Enter robotic process automation (RPA). The Institute for Robotic Process Automation and Artificial Intelligence defines RPA as “the application of technology that allows employees in a company to configure computer software or a ‘robot’ to capture and interpret existing applications for processing a transaction, manipulating data, triggering responses, and communicating with other digital systems.”4 Generally, when we think of robots, we think of a humanoid replica (think C-3PO from Star Wars), an automatic arm on a manufacturing floor, or perhaps the Roomba floor vacuuming system. Applied to software, we now refer to something a little different and a bit easier to understand: the use of sophisticated computer software that automates rule-based processes without the need for constant

human supervision. And here we have the birth of the bots – also known as chatbots. Bots have the potential to get information to users, to implement rule-based transactions, and provide services based on those rules – think making a dinner reservation online without ever talking to a live person. In the work environment, they may be similar to the “just-in-time, just enough” learning provided to an employee – for instance, while undertaking a purchase order, an employee may pause – trying to figure out the next step. The learning object, sometimes based on most frequently asked questions, pops up with: “I see you are hesitating to complete this task – do you need to know what the next step is, given the amount of this transaction?” Today, bots are more than watchful helpmates – they are smarter and they learn. For example, a company named Heek markets a bot that can help people create websites, designed especially for smaller companies that may lack an internal creative department. Its conversation may start like this:

Hello! I’m a little robot and I am going to help you create a website that perfectly fits your needs. It will be as simple as a conversation between you and me. I am called Heek, and you? Please type your name below. It then gathers information, presents editable choices, and supports options for content, colors, and artwork along the way. Bots have the potential of eliminating many of those repetitive questions that HR professionals get, and ease the ability for employees to manage transactions better for themselves. Given that up to 90 percent of the time spent on mobile devices is using email and messaging platforms, chatbots • Workforce Solutions Review • July-September 2017


have rapidly gained popularity in business. They can “converse” with employees via an application or SMS (texting). Using SMS, for example, an employee may want to request a day off in the following week. First, he needs to know if he had the vacation time to use, and second if he can take the day off. The bot conversation may go something like this: Human: Day off rules. BOT: I see you are interested in taking a day off, can I help you? Human: Sure. Bot: You currently have five days remaining for vacation before December 30. What day would you like to take off? Human: Next Wednesday. Bot: There are sufficient staffers on the retail floor in your department next Wednesday, so it looks promising. Shall I send the request to your supervisor?

questions are more likely given a particular starting point. Bots can also provide compelling data with which HR can transform policies and practices that are not working optimally. Bot-collected data lets HR know what kinds of questions are most frequently asked or what series of tasks as a combination are most likely to appear together. This can lead to better policy decisions and increased clarity in existing explanations to employees and transactional efficiency. And, once a process is reengineered and programmed into the system, it will be executed accurately across the company. Human Resources has a plethora of processes and transactional tasks ripe for automation. Some of these are being addressed by the vendor community in human capital management and talent management software. As organizations increasingly add bot use to their service repertoire, there are several steps to consider: •

Which self-service processes are best augmented with bot support?

Which processes that now are manually managed can be streamlined or totally accomplished by bots?

Bot: Your request is approved. See you back here on Thursday morning! Is there anything else I can help you with?

What data can be gleaned from the organization’s use of bots and how can it be used to improve processes and policies?

Because they are smart software, bots can be programmed to know relationships. If a parent is returning from new-child leave and requests to start back the following week, the bot can ask related questions, such as: “Do you need to add your child to your beneficiaries or to your medical benefits? Do you need to enroll in the company’s child care program?” And, not only are they smart, bots learn; they learn over time which kinds of follow-up

What technical expertise, if any, does the organization need to optimize bot use?

Human: Yup. (Supervisor – one click response approving the request on a mobile phone app)

Bots are just one instance of automated intelligence and machine learning that is gaining prevalence in business today. With the current emphasis on artificial intelligence in HR today as only the beginning, the ability to increasingly automate repetitive, transactional tasks will free professionals to address more strategic activities.

Endnotes 1

Mike Sondalini, Lifetime Reliability Solutions, Unearth the answers and solve the causes of human error in your company by understanding the hidden truths in human error rate tables.


Martin Stumpe and Lily Peng, Assisting Pathologists in Detecting Cancer with Deep Learning, Google Research blog March 03, 2017.


Martin Stumpe and Lily Peng, Deep Learning for Detection of Diabetic Eye Disease, Google Research Blog November 29, 2016.


About the Author Dr. Katherine Jones is a partner and director of research at Mercer in Talent Information Solutions. With both academic and technology industry experience, she has been a high-tech market analyst for 18 years. Her doctoral degree is from Cornell University. She can be reached at or @katherine_jones.


July-September 2017 • Workforce Solutions Review •

EIR Compensation Analytics™ provides fast, easy compensation reporting and analytics for SAP® SuccessFactors®

F  


A  

Extend the capabilities of SAP SuccessFactors with EIR Compensation Analytics

EIR Compensation Analytics™ a powerful SAP HANA® Cloud Platform extension application is now available for use in your organization’s SAP SuccessFactors Compensation instance. Provide every user access to fast, accurate, compensation cycle reporting and analytics. Contact EIR today! 855-589-9451 • Workforce Solutions Review • July-September 2017


Florida Community

Leveraging Leadership, Technology & Human Capital International Association for Human Resource Information Management

JOIn US & EngAgE

WItH tHE FLORIdA LOCAL COMMUnIty 2017 IHRIM Florida’s Reinventing Human Capital & technology Summit digital transformation - disruptive Innovation September 14-15, 2017 the Shore Club | 1901 Collins Ave, Miami Beach, FL

Join us @ this must-attend conference to engage, contribute & learn top Industry Leaders & Our Esteemed Panel will share their expertise at this exclusive 2-day event! • •

Interactive exchange, workshops, and a vendor collaboration event Leave with solutions you can use, and alliances you can depend on

Share your needs with key leaders of HR and C-level executives from multinational organizations Efficient & effective, it’s all about getting down to business!

Why the local community is important to you...

. . . . .

Get spotlighted personally & professionally among your peers & expand your Sphere of Infuence; Broad base of Alliances: increase your reach to our HR Technology community, industry analysts, consulting firms, vendors and other strategic HR alliances; Vendors and potencial customers interact closely; Executives and practitioners share advanced knowledge in Human Capital Management and enabling technologies; Make a difference for your company’s bottom line and for your career.

EvEnt AgEndA

Attend our event & earn credit hours for your HRIP Recertification:

Sept 14 | 9:00 AM - 7:30 PM

Scan Exclusive Invitation

Exciting Keynote Topics, Networking, Panel Discussion, Meet Vendors - Lunch Provided, Happy Hour Expand your network

Sept 15 | 8:30 AM - 12:00 PM Design Your Story Workshop




. . .

7.5 Credit Hours for Sept 14th Event 3 Credit Hours for Sept 15th Workshop

Since 1980, IHRIM is the HR Technology organization at the forefront of: HR IT Research HR Tech Education HR Data Analytics HR Tech Advancement

WSR July-September 2017  

Workforce Solutions Review is a peer-reviewed publication of the International Association for Human Resource Information Management, whose...

WSR July-September 2017  

Workforce Solutions Review is a peer-reviewed publication of the International Association for Human Resource Information Management, whose...