Harvard President Alan M. Garber said Wednesday he is "convinced" that artificial intelligence will drive advances in knowledge faster than at any other time in human history. The remarks, delivered on the first morning of fall classes, signal growing institutional support for integrating AI across both research and teaching at the university.
Garber's brief address at Memorial Church focused on AI's potential to compress research timelines that once spanned decades. He recounted a message from a faculty member who told him, "I've accomplished in the last six months what I thought would take the next 10 years, and built up new ideas that have been little more than dreams."
"When I think about the age now unfolding, I tend to focus on the costs, risks, and benefits of the changes to come, and what they mean for our mission and our community. I rarely think about dreams," Garber said. "But perhaps as we continue to ponder challenges and possibilities, we should give ourselves greater license to imagine the future that we hope to inhabit."
Classroom adoption gains administrative backing
The same morning, Harvard College Dean David J. Deming sent an email to undergraduates endorsing widespread AI adoption in classes. Deming encouraged not just science courses but writing-based courses to incorporate the technology. He also suggested professors were better off "getting out of the AI-detection business."
Deming, who previously warned that overreliance on AI circumvented the learning process, said he would produce advisory guidelines on AI use this year. Harvard currently has no universal AI policy, leaving faculty members to set their own classroom rules.
This stance contrasts with several peer institutions. The University of Chicago banned AI use in its core social science sequence last month. The University of Chicago Law School and the University of California, Berkeley Law School have also adopted restrictive approaches.
Leadership signals against restrictive regulation
Garber has previously described himself as an "optimist with some asterisks" about AI as a teaching tool. In a July talk, he said it was no longer possible to be at the forefront of academic research without using AI.
This summer, Deming hosted Harvard Corporation member Mariano-Florentino Cuéllar on his podcast. Cuéllar warned against "clunky" regulations that cannot adapt quickly to technological progress. He was later named chief global affairs officer at AI company Anthropic.
Even while praising AI's potential, Garber defended the value of human intellectual work. "Human intelligence is not just precious; it is distinct from any intelligence that humans have created," he said. "As astonishing as the technical advances are, learning is at the heart of what makes us human."
Why this matters for science and research professionals
Garber's claim that a colleague compressed a decade of work into six months using AI is a data point worth watching - not a proven outcome. But it reflects a pattern researchers are seeing across disciplines: AI tools are shortening the cycle between hypothesis and result. For scientists weighing whether to invest time in AI Learning Path for Research Scientists, the signal from one of the world's largest research institutions is clear. The administration is betting that AI literacy will soon be baseline, not optional. The broader field of AI for Science & Research is moving from speculative to operational, and the gap between early adopters and wait-and-see peers may widen faster than expected.
Your membership also unlocks:
