Issue 38 — The Missing Piece of Penn’s AI Strategy

Penn has moved quickly to establish itself as a leader in AI through new academic programs and AI-focused research. But while the university has invested heavily in AI expansion, it still lacks a clear plan for one of the most important questions AI raises: what students should still be expected to learn and demonstrate independently.

Penn has made AI-based research and education a clear institutional priority. In February 2024, the engineering school introduced a B.S.E. in Artificial Intelligence, the first undergraduate AI major in the Ivy League. Soon after, Wharton launched an undergraduate concentration and MBA major in Artificial Intelligence for Business, alongside a growing number of AI-focused executive education programs.

Penn Forward includes Penn AI as one of its primary initiatives, aimed at positioning the university as a national leader in AI policy and research. SAS Horizons similarly launched the AI & Data Collaborative to integrate AI into research and teaching across SAS.

These initiatives are important for Penn’s continued prominence and for serving the public good. Penn researchers are already using AI to advance medical research, scientific discoveries, and business innovation, and the university is right to invest in these efforts.

But while Penn has aggressively pursued leadership in AI research, it has not put forward the same coordinated focus on how generative AI is reshaping higher education itself.

This spring at Harvard, 40% of surveyed students self-reported using AI in ways that were “inappropriate or against class policy.” In 2024, a survey from Penn’s Student Committee on Undergraduate Education found that 83% of undergraduates had used AI to assist with their work. This has likely only increased as AI capabilities and prevalence have exploded.

During last week’s Penn Alumni Weekend celebrations, President Larry Jameson moderated a panel discussing AI research at Penn aimed at accelerating scientific discoveries and fostering learning for students across ages. During the discussion, Wharton professor Hamsa Bastani described how learning often comes through “struggling” with material rather than immediately receiving answers through AI.

The concerns around AI usage extend beyond academic dishonesty. If students rely too heavily on AI to complete their work, they risk losing the productive struggle that develops reasoning, writing, and problem-solving skills.

Traditional homework assignments, problem sets, and papers may no longer reliably test critical thinking and problem-solving skills when answers are instantaneously available. If traditional forms of assessment no longer fully measure student learning, Penn must rethink what it actually means to learn, which skills are most critical in earning a degree, and how those skills should be evaluated.

Across the country, professors are already adapting. Princeton is overhauling its exam system with new proctor requirements, and many professors across schools are moving toward in-class or oral examinations designed to better evaluate independent thinking. The Wall Street Journal recently reported that grades in writing- and coding-focused courses have risen significantly since the arrival of generative AI, raising broader concerns about grade inflation and trust in academic evaluation.

But Penn has not shown the same level of institutional urgency toward addressing how AI is changing the university classroom itself.

This week, Dean of the College Peter Struck discussed the new undergraduate curriculum with Penn Today: “We’ve got cooperative, seminar-based learning that gives us a chance to consult on best practices, to implement new innovations, to work with AI in our classrooms.”

Dean Struck’s comments hint at an important reality. Smaller, discussion-based classrooms may be better positioned to preserve learning in the age of AI. But individual faculty support for seminars is not the same as a broader university strategy for how AI will reshape learning and evaluation.

In 2023, Penn introduced its first university-wide guidance on generative AI usage, caveating the document by saying it “should not be viewed as a final policy.” The guidelines provide broad recommendations on security, safety, and academic integrity while leaving much of the implementation to individual schools, departments, and professors. That flexibility may make sense across disciplines with different needs, but it has also created fragmented rules and expectations around AI usage across the university.

The challenge is not simply individual professors preventing cheating using AI models. It is whether students across Penn are still developing the critical thinking skills that higher education is supposed to cultivate.

That question goes to the heart of higher education’s value proposition at a moment when public trust in universities has declined and employers increasingly question what grades and credentials actually measure.

No university has fully solved this challenge, including Penn. The university wants to encourage innovation while also preserving academic rigor across disciplines with very different needs. But if Penn truly wants to lead in the future of AI, it cannot focus only on research, policy, and commercialization. It also needs a clearer, more coordinated strategy for preserving rigorous learning and trustworthy evaluation in the classroom that AI is rapidly transforming.

Right now, Penn’s leadership has three months of a quiet campus to begin creating a strategy for how to preserve student learning and sharing progress publicly. There is no more time to waste.

Curated highlights from this week’s Penn news

Thank you for reading the Franklin’s Forum newsletter! We love connecting with our readers — send us your thoughts and questions, Penn news, and ideas for future issues. If you enjoyed this edition, please spread the word by forwarding it to friends and classmates.