A Penn professor used AI to replicate part of a master's course — and says it threatens universities' business model

• A Penn economics professor said he used AI to replicate a course in a master's program.
• Jesús Fernández-Villaverde said Anthropic's Claude designed the course and guided his learning.
• He said the chatbot likely rivaled most university professors in curating what and how to study.

What would normally take a week in a master's program took one University of Pennsylvania professor 12 hours — with the help of AI.

Jesús Fernández-Villaverde, an economics professor at the University of Pennsylvania, said he used Anthropic's chatbot Claude to design a personalized course on sociologist Erving Goffman, complete it, and discuss the material in real time.

After around 12 hours of study, he said he reached a level of understanding comparable to what a master's course would deliver over a week, covering roughly the same total study time.

"Goffman would typically receive one week, which translates to roughly 9 to 12 hours of total student effort," he told Business Insider.

"The time I invested is comparable to what a student in a well-structured course would spend," he said. "The ability to learn at this level, at close to zero marginal cost, is extraordinary."

AI as a personalized tutor

The process had three stages, Fernández-Villaverde said.

First, Claude generated a tailored syllabus based on his existing knowledge, including readings, key themes, and connections to other thinkers.

Second, he read the sociologist's books, including "The Presentation of Self in Everyday Life" and selections from "Asylums" and "Stigma." Claude selected the readings and structured the order, he said, "but I did the reading myself."

Third, he used AI as an interactive partner, asking for clarifications, connections, and parallels to what he was already familiar with from economics.

"It is the difference between reading a book alone and reading it alongside a knowledgeable colleague who has infinite patience and no office hours," he said of his experience with the AI-generated course.

He said Claude "excelled" at curating what to study, in what order, and what to emphasize, given a student's particular background, which he described as "one of the hardest things a professor does."

"Claude did this at a level that I would say exceeds the 90th percentile of real professors, at least for this kind of task," he added.

Not a perfect substitute

However, Claude fell short on several levels, Fernández-Villaverde said.

It doesn't challenge students in the way great teachers do.

"It answers the questions you ask rather than the questions you should be asking," he said.

It also can't replicate the peer experience of a classroom. Still, he said that human professors aren't perfect either, adding that he doesn't always answer his own students' questions precisely or correctly.

"One should compare Claude not to the ideal professor but to the real one," he said.

A reckoning for universities

Fernández-Villaverde sees AI as a major positive for learning — but a serious challenge for certain institutions.

Several professors and prominent academics, including University of Texas history professor Steven Mintz and economist Tyler Cowen, have echoed similar concerns, saying that AI is exposing long-standing weaknesses in higher education, including standardized assignments and outdated teaching models.

For Fernández-Villaverde, the implications are more economic: AI is forcing universities to justify what students are paying for.

University programs built primarily around delivering lectures, he added, are especially exposed.

"If your main value proposition is transmitting existing knowledge in the classroom, and a student can get a comparable or better version of that for $20 a month, the business model is under severe pressure," he said.

However, that doesn't apply equally across all institutions, he said, adding that top universities still offer advantages AI can't easily replicate, including proximity to the research frontier, strong peer networks, and valuable credentials.

While he said he doesn't think AI will eliminate traditional higher education, he believes it will force them to rethink what they offer.

"The universities that will thrive are those that offer something AI cannot," he said, citing research mentorship, laboratory access, genuine peer communities, and a credible credential.

"The ones that are essentially selling access to lectures and a diploma will face the hardest questions," he added.