Yale University does not have a single policy on artificial intelligence in the classroom. Instead, professors and departments are creating their own rules for how students can use AI tools.
The Yale Daily News reported on September 28, 2026, that approaches vary widely. Some faculty ban AI entirely, while others allow limited use.
Large language models have grown more capable since ChatGPT launched in November 2022. Anthropic, the company behind the AI assistant Claude, says on its website that the length of tasks AI can complete on its own has been "doubling roughly every four months."
Departments Take Different Approaches
Ben Glaser is director of AI initiatives in the humanities at Yale's Poorvu Center for Teaching and Learning. He works with faculty across departments to build AI policies.
"It's an ever-changing technology," Glaser said. "It's better to have a conversation about specific guidance rather than assuming."
Glaser said clarity and transparency matter most. He wants professors to explain the reasons behind their policies and any cases where students may use AI.
In the Mathematics Department, professors have given less weight to work done outside of proctored settings. John Hall, a senior lecturer in mathematics, said exams make up most of the grade in his course.
"I'm the course director for multivariable calculus. 80 percent of the grade is on the exams, and that's pretty typical," Hall said.
The Philosophy Department has less agreement. Assistant professor Lily Hu said faculty have discussed AI many times but have not settled on shared policies.
Strict Bans and Flexible Rules
Hu does not allow AI in any part of her courses. That includes written work, presentations and even brainstorming.
She said learning involves "strain and confusion and discomfort." In her view, students who use AI to skip these steps are not truly learning.
"These are not unfortunate hurdles that happen to accompany the learning process, they're the key experiences of learning itself," Hu said.
Julian Posada, assistant professor of American Studies, takes a different approach in his course "Platforms and Cultural Production." Students there may use AI for brainstorming, clarifying ideas or spotting grammar and style problems.
Posada does not allow AI to generate or rewrite language, analyze arguments or read assigned materials. He said he opened up his policy after students, mostly seniors, told him they used AI heavily during summer internships.
Hall's MATH 1200 syllabus lets students use AI as a study tool but not to complete problem sets. He encourages students to use peer tutors and office hours instead.
Hall said he might promote AI tutoring at a school with fewer teaching assistants. "But at a school like Yale, where there are so many resources, so many people there to help, I don't want to do that," he said.
Glaser said many instructors are promoting in-person interaction in the spirit of a residential college. He said students and faculty value experiences that AI can sometimes get in the way of.
The new classroom policies come as AI tools keep advancing. OpenAI released its newest large language model, GPT-6 Astra, on September 3, 2026.
