AI Takes on the Law

The legal field is arguably one of the more difficult challenges to Artificial Intelligence, if not the most boring for most humans. AI is now tackling that challenge head on with a number of startups.

These intrepid startups (is that a tautology?) are taking on specific challenges in law such as reviewing contracts to identify risks, provide legal analytics, predict the outcomes of court cases, search for information that supports or opposes the positions of the litigants, assess relevance of documents for a case at hand, monitoring and protecting intellectual property, and so on.

In the distant past (well, about four or five decades, which counts as distant by the standards of technology), expert systems were one of the more predominant areas of AI. The pioneering examples include Mycin in the mid-70s for diagnosing infectious diseases and Dendral for identification of organic molecules. One of the big challenges to designing effective expert systems was knowledge elicitation: the problem of identifying the topmost domain experts and getting them to articulate their expertise.

In the legal field, the goal is probably less grandiose than that of the original expert systems, but impressive nevertheless. Startups in automating legal expertise seek to deliver legal knowledge, access correlated information, and automate the drafting of legal documents. The problem is less of knowledge elicitation from experts and more of examining an enormous number of documents and gleaning insights using data analytics and machine learning.

The legal profession is very conservative, preferring to stick to the status quo. These startups will have their work cut out to persuade the lawyers to let go of their fears and embrace AI.