Why Legal Research Should be Automated

Lawyers work too hard. The average lawyer clocks 50+ hours a week, and 75% of lawyers admit they work outside of regular business hours. Something's gotta give; either their home life, the quality of their work, or their mental health. Part of the problem is that lawyers are still dabbling in work that can be automated. Legal research is one of those areas. With database search tools, finding the right cases to apply to a given file is like trying to find a good show on Netflix. You're likely to sort through a lot of irrelevant content, and watch too many trailers for shows like Cupcake Wars and The Secret Life of Broccoli before deciding to give up and go to bed. The process of hunting down the right cases, and trying to digest them, can take anywhere from 2-6 hours for a single research issue.

Lawyers have one of three choices:

First, they can bill their clients for the time, and likely receive an angry phone call asking why they should pay $3,000 for their lawyer's self-education. (I pay you to know the law! Not to look it up!)

Many lawyers, then, prefer to avoid upsetting their clients with a surprise bill that's higher than their monthly mortgage payment. So they choose to write off some, or all, of those hours spent on research. Which means they just logged 2 billable hours instead of their 8-hour quota that day.

Some lawyers decide to dodge this problem by just not doing the research. Maybe they're 60% sure they know what the law is, but they're comfortable with that. But how would that lawyer feel if their doctor told them I'm 60% sure I know how to do this surgery, but don't worry... you're in good hands. Lawyers should be 100% confident they know the law, or risk losing their client's confidence.

What's the solution to this problem? Legal research should be automated. Lawyers should spend 5-10 minutes to know the law, not 2-6 hours. A.I. can help with that. A lot of the pushback against A.I. doing legal research focuses on the perceived lack of accuracy. But A.I. has the potential to be far more accurate than humans. IBM's Deep Blue beat Gary Kasparov at chess. Watson diagnoses lung cancer with 90% accuracy, compared to 50% accuracy for human doctors.

Tesla's Autopilot A.I. has an accident rate of 1 for every 3 million miles driven, while the human equivalent is an accident every 450,000 miles. For legal research performed by humans, the question of accuracy assumes that research is a science, and not an art. This is a mistaken assumption. If you give the same legal question to two different lawyers and ask them to draft memos answering it, those memos will almost certainly differ.

Human advocates inject subjectivity into the equation by trying to find cases that support their client's position, while ignoring cases that hurt their client's position (either consciously or subconsciously). A.I., on the other hand, doesn't care who wins the case. It can be purely objective about the law, and if a case is relevant, it will make it into the memo. If it's not, it won't. A.I. can make legal research less of an art, and more of a science. In summary, A.I. can achieve massive time-savings, and greatly increase accuracy for legal research, and it should be automated.