UC Berkeley Law has adopted a new policy that prohibits the use of AI in student academic work submitted for credit.
getty
The University of California, Berkeley School of Law has adopted new, stricter policies governing the use of AI by law students. The rules, which go into effect this summer, will ban the use of AI in most activities related to academic research submitted for credit.
The new policy states:
- “The use of AI to assist in conceptualizing, outlining, drafting, revising, translating, or editing work submitted for credit is prohibited. The use of AI is prohibited in any exam context or for any purpose. Students may not upload course materials, including assignments, readings, slides, class notes, or other class content to the Generating AI system. may be used for the limited purpose of identifying sources such as case law, statutes, and secondary sources. Students are responsible for the accuracy of the research and all other aspects of the submitted work. Citations to non-existent sources give rise to the presumption of prohibited AI use.”
Potentially broad exceptions are made for courses that are “intended to teach AI fluency” and other courses where instructors decide to seek written permission to deviate from the default prohibitions for pedagogical reasons.
In a series of posts yesterday about
While the previous policy allowed students to use AI technology to help revise required writing assignments, the new policy specifies several prohibited uses of AI, including:
- Ask an AI tool to brainstorm a paper topic or paper (no conceptualization)
- Have an AI tool suggest the organizational structure of your paper (no outlines allowed)
- Have an AI tool create a text that summarizes the legal rules to be used in the paper (drafting is prohibited)
- Use AI tools to identify and cut duplicate parts in papers (modification prohibited)
- Have an AI tool correct grammar mistakes and refine your paper (editing prohibited)
- Let AI generate an exam summary and use its elements in the exam (exam use prohibited)
- Have AI translate papers written in other languages into English (translation prohibited)
While students can still use AI to teach themselves and prepare for classes, Hoofnagle says the main concern is that “in the classroom, we don’t want students to write the best paper possible; we want them to write the best paper possible according to their abilities.”
UC Berkeley’s new rules will almost certainly attract the attention of other law schools. In addition to its generally excellent reputation as a law school, the school is also a pioneer in education and research on the use of AI by lawyers, lending further credence to concerns about the appropriate and inappropriate role of AI in legal education. In fact, the rationale for the ban applies to higher education and the type of learning that higher education should promote among students in general.
This policy acknowledges that future lawyers will need to be fluent in the use of artificial intelligence. In fact, most lawyers and many large law firms use it for a variety of purposes. However, the rationale for this rule is that good lawyers “need to combine the use of AI with the cognitive skills necessary to strategically deploy technology, critically evaluate its output, and uphold ethical obligations to clients and the legal system.” thought It remains a prerequisite for a good lawyer (and quality legal education). ”
As AI technology matures and its use increases, other types of prohibitions and penalties are being introduced to maintain the integrity of research and scholarship activities. Preprint platform arXiv, for example, has a code of conduct that imposes an immediate one-year ban on authors if it finds “indisputable evidence” that a submitted paper contains “bad language, plagiarized content, biased content, errors, misrepresentations, false references, or misleading content” written by a large-scale language model.
“If a submission contains indisputable evidence that the authors did not check the results of the LLM generation, this means that nothing in the paper can be trusted,” argued Thomas Dieterich, chair of arXiv’s Computing section. Post to X. In addition to the one-year ban, authors found to have submitted papers containing AI-generated errors or hallucinatory references will be required to have their subsequent arXiv submissions “first accepted by a trusted peer review venue.”
