Not with flashy new majors or gadgety electives, but by stripping it all back to the one question that matters: What can humans still do better?
Turns out, quite a lot. But we’re not teaching it.
Some schools are starting to. Quietly, almost experimentally. Like Northeastern’s co-op program, where students embed in industry and bring back AI-assisted workflows into classrooms that barely acknowledge them. Or like ASU, which partnered with OpenAI to prototype GPT-4 as a TA, a study partner, and—unofficially—a barometer of what AI gets wrong.
One business school tried a pilot last spring: replace the final exam with a scenario. A simulated market collapse. Students could use anything—AI, financial models, competitor data. What mattered was how they reasoned through chaos.
The strongest teams didn’t have the best models. They had the best disagreements. They fought clean, adapted fast, rewrote assumptions. One team used an LLM to test its blind spots. Another rewrote the AI’s prompts mid-simulation after it started hallucinating trends.
They didn’t need to outthink the machine. They needed to out-human it.
“The purpose of education in an AI age is not to outpace the machine, but to outmatch it—by being more human.”
—Joseph Aoun, President, Northeastern University
Minerva dropped lectures. Students now learn through live debate, design sprints, and cross-cultural problem-solving. Olin engineers spend year one building—not just learning—and are forced to reckon with the ethics of what they make. Arizona State has let GPT-4 loose in classrooms, where it tutors, nudges, even challenges.
These aren’t edge cases. They’re sketches of the next normal.
One history major at Minerva had to rewrite a Cold War policy memo from the Soviet perspective, using AI-generated counterfactuals as raw material. She didn’t just analyze the past. She argued with it.
Another student, this one at a community college in Ohio, used Midjourney to reimagine public health posters from the 1950s for a visual rhetoric class. Her professor admitted she didn’t quite know how to grade it—but also, that she didn’t want to go back.
The change goes deeper than new tech. It’s structural. The walls between disciplines are cracking. Curricula are bleeding into each other—data science meets journalism, ethics wraps around computer science, psychology informs machine learning. Students don’t major in silos. They learn in systems.
And they all need AI literacy. Every last one of them.
“AI should be taught like writing: a core skill, not a technical specialty.”
—Fei-Fei Li, Stanford
Because what AI reveals, under all the math, is whether we still know how to think. Not regurgitate. Not optimize. Think—ethically, critically, under pressure.
It also demands something more elusive: that we remain unmistakably human.