Earlier this month I was watching The Daily Show with Trevor Noah (an American late-night talk and news satire program that often raises important issues in humous ways), and saw this segment:
Ronny Chieng (one of the senior correspondents) goes on to explore artificial intelligence (AI) in the legal industry and asks the question – does AI belong in the legal industry? With recent advancements in AI, it seems like there is nothing AI can’t do, and the legal field is no exception.
A primer however is to note the AI technology that deals with legal information is natural language processing (NLP) or computational linguistics. To date, firms have been using NLP to process paperwork faster and avoid potential bias in the legal industry. More recently, people in the legal industry and beyond have been getting curious if AI can do more than just process paperwork.
So, can we use AI in the legal industry? Should we? And can AI act as a judge and pass a verdict?
On one hand, it could be argued that AI would be a good judge as it does not have emotions, so it could be a useful tool in reducing bias and potentially even corruption when reaching a verdict. However, our judicial processes and the laws that we have created are human in nature and unfortunately, a lot of these laws have historically been used to improperly prosecute minority groups. This means that while we could teach AI to understand our laws and perhaps even reach a conclusion, it would still be susceptible to the inherent bias these laws possess as they were created by us, humans.
This also means that AI is only as ethical as we are. So even though it could hypothetically take into account human attitudes when passing judgement because the laws that we feed it were created by us, humans, this might not be a good thing as we have a poor history of making ethical decisions.
Moral crimes or criminal law requires understanding of ethics that AI does not have. However, I would consider using AI in more objective disputes that are tied to property rights (Coase theorem) and maybe for damages (e.g., insurance claims).
Because most so called "AI" models are distributional, meaning that they do not reach a right or wrong conclusion, they report results such as “40% Yes”, “30% No”, and “30% Uncertain”. This means that even if we did try to get AI to resolve cases around moral crimes or criminal law, these issues would be too complex for it, as they would require decisions that are more nuanced than a simple “Yes”, “No”, or “Maybe” in percentages. It’s literally a gray-area! However, AI or more specifically, natural language processing technology, can help aid in the legal interpretation of such disputes by:
1) Helping the lawyers save time by identifying precedent.
2) Help gather all the necessary facts and organize them.
3) Detect potential judicial bias or bias in the response of the jurors.
In addition to the ways AI could be use in the legal industry that I already mentioned above, I also think that AI could be used to support payment collection as it would be an option for cities or municipalities with limited public lawyers available. AI could save resources by identifying people that are skipping child support payments and are statistically most likely to pay, and therefore protect children and their parent/guardian.
All in all, I am excited to see the recent advancements in legal AI. However, I don’t think that AI is ready to become a judge and as flawed as humans are, we can still make better judges that a machine.
If you are curious about this topic and would like to learn more - check out the resources below.
Comments