University of Chicago Law announced a new AI policy, seeking to combine incorporating AI in some areas while making other areas “AI-resilient.” The latter includes an in-person oral presentation for upper-level writing courses, prohibiting laptops and tech in 1L classes, and in-class/no-tech exams in 1L classes.
I have prohibited laptops in all my classes for years, so I support the second plan. I had not considered the connection between AI and in-class computers–I guess students could use Claude or ChatGPT to get answers in the Socratic dialogue in the moment, although it would seem too fast-developing. But to the extent AI resilience pushes in the same direction as the pedagogically beneficial removal of laptops, awesome.1
The third point highlights what I hate most about AI in legal ed: It has forced us backwards on assessments. Take-home assignments–which allow for better-written, better-edited, better-analyzed, more thoughtful essays written in a context that better reflects the reality of legal work–are ideal, but also prime targets for AI abuse. We return to time-crunched in-class assignments that do not reflect the real world and do not allow for the same thoughtful and complex analysis.2
The answer may be more oral assessments. Josh Blackman describes the oral mid-term he wants to offer. I have described my use of oral arguments as the final assessment in my upper-level Fed Courts and Civil Rights classes. I could try to recreate something similar for Civ Pro and Evidence, although class size makes it more difficult.
Another thought is assigning greater evaluative weight to portions of class participation. My Evidence class is built around two fictional cases and evidentiary questions from those cases; during class, we go through the rows asking students to argue the evidence from each side and as the court. In essence, each student does 3-4 short oral presentations (similar to what Josh describes) over the semester. Perhaps I could increase the value of each of those. Not sure how to match it in other classes, which are less problem-based and more Q&A dialogue.
- This faces the problem of the recent increase in laptop accommodations. ↩︎
- IN Civ Pro and Evidence, I emphasize properly numbering rules, getting to the precise provision in the right form (e.g., FRCP 26(c)(1)(C)(iii)). I think this is important in code classes, part of learning the “language” of the code. I cannot expect that precision in three or four hours. ↩︎