According to the 2026 State of the Academy (SOTA), 88.3% of students use Artificial Intelligence (AI) for some academic/educational purpose, chief among them to explain concepts, create study guides or practice questions, and summarize texts or information. As AI becomes increasingly present in student work, faculty across departments have begun reassessing how to harness these tools in the classroom.
The Natural Sciences Division recently held a day-long workshop with Hanspeter Pfister, An Wang Professor of Computer Science at Harvard University, who led faculty in reflecting on AI’s productive uses, limitations, and implications for students. Chair in Physics John Rogers described his main takeaways.
“We have to really be as proactive as we can be in figuring out what [AI] means for our students and ourselves because, whether we like it or not, it’s here and it’s transforming the way that many things are done. We don’t really have the luxury of ignoring it. Another big idea was that we want to… create assignments where if it’s going to help students to use AI that we’ve built that into the assignment so that everyone gets equal benefit from it,” said Rogers.
Brian Faulk, Head of the Natural Sciences Division, elaborated on a point from the discussion that the Division hopes to explore next year.
“One of the things that Professor Pfister encouraged us to do is create AI-literate graduates, or at least that’s something that Harvard feels strongly about, because it is going to be a skill, much like using email and navigating the internet was when I started going into the workforce… [Creating] students who are AI-literate, who are using it to support their learning but not replace the process of learning,” said Faulk.
Nick Zufelt, Instructor in Mathematics, Statistics, and Computer Science, described his approach to AI and how it varies between his math and computer science courses.
“In my math classes, I would say that AI is available, but it’s not really talked too much about. I told students that they can go ahead and use Khanmigo, [or] they can also use something like ChatGPT. Obviously not on a test. That’s where the vast majority of the grade comes from in my math class, so that’s why I feel relatively comfortable allowing students to use it. I’m going to start incorporating a little bit more AI by showing these students how I would tackle math problems using AI. That’s probably coming at a later stage,” said Zufelt.
Zufelt continued, “In my computer science classes, we use AI a ton. There are certain [tasks] that I’m comfortable with students allowing AI to perform. For example, it could be the case that AI is used to do research on some computer science concept. So maybe you are writing the code to have a character move across the screen in a game. You can do research on the side and have AI explain to you how something works. I think a lot of people, especially a lot of teachers, worry about what we would call hallucination, or confabulation, where AI just makes up stuff that’s not true. The cool thing about computer science is you can just run that code and test with your own eyes if it’s working or not. So that’s why I feel relatively comfortable in computer science, allowing AI use, as long as students are being responsible with it.”
Keri Lambert, Chair in History and Social Science, shared that humanities instructors have used AI-powered tools to better understand their potential. She expressed her hopes for AI literacy to complement integration in the future.
“If instructors use AI at all for work, it’s mostly to understand the tools better. For instance, some instructors have tried to learn how ChatGPT analyzes historical evidence, how Gemini produces research notes, or how Claude organizes historical arguments… In the next couple of years, I hope History can collaborate with the other Humanities departments to figure out how to promote better AI literacy so that students can more critically evaluate AI tools as well as the content they generate,” wrote Lambert in an email to The Phillipian.
Within the department, AI usage has varied among instructors. Eric Denby, Instructor in History, explained how AI has enabled him to expand his creative range in the classroom.
“I have found AI useful in creating historical simulations that I use during group activities. While it doesn’t necessarily lessen time, as I double check all of the information for accuracy, it does allow me to explore different ways to make history come alive in my classroom. As far as workflow, it has not reduced much time in lesson plan, though it has become a nice brainstorming partner,” wrote Denby in an email to The Phillipian.
Lambert also expressed reservations about Khanmigo, an AI tutoring tool that Andover piloted this Spring Term.
“The Khanmigo writing tutor tool did not offer helpful advice in the demo that I saw; in fact, it affirmed really poor writing, so I worry it could mislead our students… [We’re] trying to help students learn how to think, whereas I worry Khanmigo incorrectly implies History as a discipline is more in the business of teaching students what to think,” wrote Lambert.
Many students have noticed these shifts in usage and enforcement firsthand. Nafi Diagne ’26 shared her observations of cultural attitudes towards AI during her years at Andover.
“Before my Lower year, nobody really knew what it was. Before my Upper year, no one knew what it was. I heard the word ChatGPT three times. Upper year, all [of a] sudden it was this huge thing. The teachers, especially in the History Department this year at least, have talked a lot about how checking for AI infringements… has taken so much time, and they shifted from absolutely not using it at all to giving you guidelines around it, just because they knew students were using it a lot… They started doing, in my English 300 [class], no out-of-class essays, whereas in my earlier years, all my essays were out of class,” said Diagne.
Kami Zheng ’27 explained how AI integration has varied greatly across her classes, particularly in her teachers’ viewpoints.
“My past English teacher, I feel like he felt personally attacked by AI. I understand that because it’s an insult to [teachers’] work. Some teachers are so against it that they refuse to interact with it at all. And then there’s some teachers that don’t enforce anything about AI. For my teachers, I’ve definitely seen some of them become more strict with it, but not necessarily let us use it more,” said Zheng.
Felisha Li ’28 noted a similar increase in enforcement around AI. She highlighted the different usages of AI and their varying implications, citing an example from her history class.
“I’ve noticed a lot more in-class essays in the English, History, and RelPhil departments,” said Li. “[Finding sources using AI is] okay in certain cases. For example, in my history manifesto project, our teacher let us use some AI to look at manifestos because it’s a gigantic database over thousands of manifestos. So it lightens the workload over something that’s tedious and doesn’t help you learn.”
Frank Hu ’26, who has begun using AI tools more frequently for his academic work, said he sees AI as a technological shift Andover will need to adapt to.
“For academic institutions, it’s a lot easier to do well and it’s a lot easier to do more. So it’s an arms race of who can use it the best in order to maximize their time… because the expectations change based on the tools that you have. I assume that research was a lot more difficult when the internet wasn’t around and in the same way, putting out a high volume of work was harder when we didn’t have AI. It’s neither good nor bad. It’s just different, and we still have to wait and see how this changes the landscape of a school,” said Hu.
