Artificial Intelligence and Policy in India, Volume 2 (2021)

Page 92

Artificial Intelligence and Policy in India

91

the case allows more than one correct interpretation (Baker, 2011). Therefore, a distinction needs to be made between plain case and hard cases. Plain cases would allow the AI, functioning as black box, to decide cases on its own. In order to balance the lack of explanation, a subsequent decision by a human judge would invalidate the AI’s decision. However, when it comes to hard cases, there is more than one correct solution as there might be multiple interpretations and therefore, the reasoning matters the most. Regarding hard cases, a humangenerated decision that provides justification is worth more than a machinegenerated decision that lacks explanation, even if it is more likely to be correct.

Solutions and Suggestions Over the past few decades, with the advent of AI, distinct use cases have been proposed. They can broadly be classified into two categories: one which is litigant-focused (to help them figure out the system and make informed and stronger claims), and one which is court-focused (to increase its productivity and transparency). This subsection makes recommendations and possible solutions as to how AI can be incorporated into the justice system in the Indian Context by analysing pieces of literature and what they propose. Guide the litigant and help them figure out how the laws applies to them. It is proposed that AI can help understand whether a given case is in accordance with the law or not, and help them conduct effective and targeted research as to how it applies to them (Cornell, 2018). Determining the latter, i.e. the question of how law applies to them is a complex task since the question of law might be straightforward or qualitative. If a question relates is easy i.e. can be answered by algorithmic calculation (for e.g. whether a person broke a traffic rule), it can be answered by a little technical challenge (Tito, 2017). If a question is complex, such as determining the intent of a person behind an offence, while mere algorithmic programming won’t suffice, it might be conquered and answered by the help of neural networks and deep learning techniques. It is also proposed that chatbots are used to answer queries of the litigants and to make them more aware of their rights (Cook, 2018). Determine the credibility of evidence and quality of the claim. Some exploratory work has highlighted how AI could automatically classify and break down the chronological events in a case, in order that it could be understood computationally (Legal Docket-Entry Classification: Where Machine Learning stumbles). Machine learning could thereby find patterns in claims to indicate whether something has been argued well, whether the law supports it, and evaluate it versus competing claims. Provide free legal aid by substituting lawyers. This vision of AI intends to develop the AI-based justice system holistically by


Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.