Opinion | Teach Students to Think With, Without and About AI

By teaching students in our universities to think without machines and then to think with them, I see an opportunity to form a generation that will advance human knowledge and well-being faster than any other.

Doomsday warnings about artificial intelligence are replete with claims that AI-enabled cheating is pervasive, that students will soon abandon college en masse, and that for those who remain, rampant credentialism will quash the liberal-arts spirit. These concerns, though real, miss the bigger, more important story about how to educate students for a world abundant with AI.

Something different and far more optimistic is emerging at the University of Chicago, including a new, AI-adapted core curriculum. Campuswide work on this topic commenced in the fall of 2024, with the creation of an interdisciplinary group of faculty to consider reforms. The report of this working group provides a simple, principled framework that will guide our changes in curriculum and classrooms. The first goal is that our students should learn during their time with us to be ethical, skeptical and ambitious about the role of AI in their education and their lives. The second is that all students at UChicago should learn to think both with and without machines.

Here, we hold fast to our belief that a core curriculum is the foundation for a college education that supports a life enriched by a love of learning and ideas. In the past, anyone proposing a change in our core curriculum would have to be steeled for a decadelong debate. Yet the urgency of this moment is prompting beneficial changes in tandem at a pace commensurate with the technological transformation.

Our core curriculum—which students largely complete within their first two years—consists of humanities and social science, as well as the study of mathematics and biological and physical sciences, and the study of civilization and the arts. Soon all undergraduates will be required to take a class in writing, without the aid of any machine, and engage in the true struggle. Real writing. Blank page. Tough comments from instructor and peers. The humanities and social-science cores will also exclude technology from the classroom.

In line with the principle that it is important for students to learn how to think with machines, there is broad internal support for the notion that to be an educated person today and in the future, one must be able to use computational tools and concepts within one’s domain, and to reason critically about how computation, data and algorithms shape what can be known and built. Likely, this will result in a new core curriculum requirement for coursework in computational thinking within the next year.

The idea that learning to use AI should be an integral part of learning how to think increasingly cuts across the whole of the university. Earlier this summer, our law school declared that first-year students will be required to learn to write and reason without technology in the classroom or the use of AI. This will be followed by deliberate and thoughtful integration of AI tools and methods in years two and three.

Law students educated in this way will be grounded in ethical and principled reasoning and adept at navigating an era when AI tools stand to shape changes in the practice of their profession. They will be prepared to lead and help shape the policy and legal frameworks that we will need to govern a democratic society where AI is widely deployed.

A challenge as serious as the disruption of how classrooms and assignments function and how students learn is important and will take time to address. In my own lab, I experience the speed and depth of this disruption as new methods change the nature of the questions we can ask—even as issues of authorship, ethical use and what is beautiful are contested.

Adapting the teaching framework of American universities will be a noisy and fraught affair. That’s a good thing. One of the greatest strengths of our universities is that they vest decision-making on curricula and course policies with the faculty, and they, in turn, inevitably have a myriad of viewpoints, each searching for a better way to teach. I am confident we will see strong outcomes from our approach at UChicago as long as we remain true to our core principles.

There are plenty of critics who insist there is little or no proper use of AI in the university, since it enables cheating and robs students of the challenge of truly learning. There are others who see AI as an integral part of essentially all curricula. The thoughtful approach emerging at UChicago understands this not as an either/or problem, but as one for which the sequence of intellectual and social development should dictate the incorporation of AI after building a strong foundation of human-centered thinking and learning in community.

What will the next wave of great minds educated in these ways be capable of? I think it will be astonishing and inspiring.

Mr. Alivisatos is president of the University of Chicago.