A while back, a few of us had dinner with professors from Indiana University’s Luddy School of Informatics, Computing, and Engineering. The professors have been instrumental as we build out and expand our expert evaluation, data and AI services.
During dinner, the topic of how AI use is handled in college classrooms came up. One of our interns (now a recent law school grad) said that in his first year, use of artificial intelligence at the law school was nearly forbidden and that in his second it was begrudgingly acknowledged that modern lawyers are using it, but there was still an anti-AI use bent.
Law schools’ aversion to use of AI is not surprising. For a different take, I asked the computer science professors how they handled it, because if there is one thing current artificial intelligence tools excel, it is coding. In fact, it seems that engineers opting out of AI use are at a significant disadvantage. Both professors acknowledged they use it in their own work and AI augmentation of coding skills has become a core piece of their curriculum.
But…and here is the point that fascinated me… the professors told me that when their students are graded, it is during a session with a teaching assistant who monitors the use of AI tools in the student’s software development workflow. During the test, the student must explain to the TA why AI was used and the student must demonstrate that he or she understood the reasoning behind the coding decision and the functionality of the code used.
There Is Knowledge and There Is Understanding
To sum up, one of the professors said to us, “there is knowledge and there is understanding.”
I immediately thought, this is a perfect pedagogy for law school. Modern law students should use AI responsibly so they are proficient with it when they go into practice. In fact, rather than remaining in a state of denial, law schools should help students learn technology skills as part of the legal curriculum.
The Risk of Blind Reliance
Obviously, if students rely on AI to generate legal writing assignments, research case law, review contracts and the like, they could lose an opportunity to learn. Blind reliance on AI is part of the reason there are many in law that still frown upon its use (see, e.g. the ridiculous frequency of false case cites in legal filings).
A Modern Socratic Method for AI
But, what if law schools adopted teaching methods used by computer science professors? Permit law students to enlist the help of AI with assignments, but then quiz them as to why they relied on what they did.
Kind of a modern day take on the Socratic method.
The professors’ use and testing of AI for their computer science classes intrigued me enough to look a little deeper into how law schools are handling AI. Some are banning its use and others are embracing it. I obviously support the latter view and was able to share some of my thoughts on the subject in a recent ABA Journal Mind Your Business column, Can law schools learn from computer science programs in the debate on generative AI?
Curious How Legal Expertise Shapes AI Models?
This dinner conversation grew out of Percipient’s work with computer science researchers on AI evaluation. Our attorneys and legal professionals provide the expert review and RLHF data that helps AI models reason like lawyers not just pattern match like generalists.
