NomDeWayne

@sightlyfixated I take the point but/and I assure you that not adding to workload is already a priority because handling this is a ton of work and a ton of especially unpleasant work for instructors. Imho the primary issue re: students is not the additional work time but the questioning of their

Law schools need to carefully consider whether AI bans can be enforced fairly. What "markers" of AI use do you expect to see, @hoofnagle? A lot of research suggests that both machines and humans are inaccurate and biased when identifying AI created content. See MIT Management, AI Detectors Don’t Work. Here’s What to Do Instead, mitsloanedtech.mit.edu/ai/teach/ai-de…. For example, nonnative speakers of English are more likely to have false positives (AI detected when AI was not used). See Andrew Myers, AI-Detectors Biased Against Non-Native English Writers, Stanford HAI, hai.stanford.edu/news/ai-detect…. AI detectors have performed better in some tests. See Matt Robinson, Do AI Detectors Work Well Enough to Trust?, Chicago Booth Review, chicagobooth.edu/review/do-ai-d…. But also consider that AI humanizers reduce the detectability of AI writing. See Microsoft, The role of undetectable AI humanizers, microsoft.com/en-us/microsof…. Additionally, many of the uses of AI that schools are attempting to ban won't be associated with submitted text to be analyzed (e.g., bans on AI brainstorming). Many people think they are good at identifying AI written text, but many research studies suggest they are very bad at identifying AI generated content. See Alexandra Fiedler and Jörg Döpke, Do humans identify AI-generated text better than machines? Evidence based on excerpts from German theses, sciencedirect.com/science/articl…; Adam Cheng et al., Ability of AI detection tools and humans to accurately identify different forms of AI-generated written content, pmc.ncbi.nlm.nih.gov/articles/PMC12…. Expert users of LLMs might fare better. See Jenna Russell, Marzena Karpinska, Mohit Iyyer, People who frequently use ChatGPT for writing tasks are accurate and robust detectors of AI-generated text, arxiv.org/abs/2501.15654. Professor skepticism of students who discuss cases not in the casebook illustrates another significant problem. Students learn about relevant cases from other academic work, including other courses and research papers, not to mention internship experiences. Why put students in the position of worrying that they can only cite cases in the course casebook?