Harvard’s latest faculty survey found that more than 60 percent of respondents viewed artificial intelligence as having a somewhat or very negative effect on their courses. Conducted from April 29 through May 22, 2026, the survey received 463 responses from more than 1,400 faculty members in the Faculty of Arts and Sciences and the Harvard John A. Paulson School of Engineering and Applied Sciences. The response total included 292 complete surveys and 171 partial ones, according to The Harvard Crimson.
Rules differ by course
The survey describes a campus without a single approach to student use of generative AI. Some respondents said their courses had no stated policy, while others allowed the tools either without limits or under varying degrees of restriction; some prohibited them altogether. The distribution was broadly similar to the previous year, as reported by the Crimson. Harvard’s Office of Undergraduate Education requires instructors to tell students how AI may be used and to post those rules on Canvas, while leaving the substance of the rules to individual courses.
Harvard is changing assessment and grading
The survey findings come as Harvard undertakes changes intended to address academic standards. Harvard reported that more than 60 percent of undergraduate grades awarded in 2025 were A’s, and faculty surveyed by the Crimson largely agreed that grade inflation is a problem. In May 2026, the Faculty of Arts and Sciences voted 458 to 201 to limit A grades to 20 percent of a course’s undergraduate enrollment plus four additional students. The change is scheduled to take full effect in fall 2027, although many courses will begin applying it this fall, according to FAS Current.
A previous grading response
Harvard has tried a major structural response to grade inflation before. According to The Harvard Crimson, faculty switched from a 15-point to a four-point GPA scale in 2002 amid concerns about grade inflation, but reported grades had risen again by the 2002–03 school year.
The same vote approved using average percentile rankings, rather than grade-point averages, to calculate internal Latin honors and academic prizes beginning in the same year. Those calculations will not appear on student transcripts. Faculty respondents also expressed support for limiting students’ use of electronic devices and the internet during class, while a majority said students do not give coursework sufficient priority.
Administration continues to expand AI guidance
Harvard has not responded to faculty concern by abandoning the technology. Harvard College introduced an AI chatbot for incoming students’ course planning and added an AI component to first-year Expository Writing. The Faculty of Arts and Sciences ended its institution-wide ChatGPT Edu pilot and said it would offer access to Anthropic’s Claude instead, according to the Crimson. University guidance also bars faculty and staff from entering confidential research or student-record information into publicly available generative AI systems.
The Harvard findings fit a wider pattern of faculty concern. A November 2025 survey of 1,057 U.S. college faculty members by the American Association of Colleges and Universities and Elon University found that 95 percent expected generative AI to increase students’ overreliance on the tools, according to the organizations’ report. Turnitin separately reported that 19.4 percent of U.S. higher-education submissions in its October 2025-to-April 2026 analysis had an AI-writing score above 80 percent, a rate higher than those recorded in the United Kingdom and Australia, according to Turnitin.
