How Penn is reimagining research in the age of artificial intelligence

In a sunlit conference room in Amy Gutmann Hall, a group of researchers convenes for a regular lunch meeting. Physicists converse with linguists, computer scientists collaborate with chemists, and psychologists speak to engineers. These postdoctoral researchers represent the University’s response to a shifting research landscape prompted by recent advances in artificial intelligence (AI).

Pointing to those gathered, Bhuvnesh Jain, who co-directs the School of Arts & Sciences’ Data Driven Discovery Initiative (DDDI) notes that “at Penn, we have a remarkable opportunity to build on our closely knit community of researchers in dozens of fields across all our Schools. Postdoctoral fellows are at the forefront of our research, and so that’s the group we are bringing together.”

This breadth is precisely what Penn is leveraging as AI development increasingly shifts from academic labs to corporate campuses, says Penn Integrates Knowledge University Professor René Vidal, who directs the School of Engineering and Applied Science’s Innovation in Data Engineering and Science Initiative.

Jain, a cosmologist by training, teamed up with Vidal, whose work centers around machine learning, to create AI x Science, a postdoctoral fellowship that brings together researchers from across the University in seemingly disparate domains to enhance and accelerate discovery. The selected fellows receive a small stipend, research funding, and dual mentorship, plus structured opportunities for peer engagement through weekly lunches, mixers, co‑hosted seminars, and hands‑on access to campus research centers, laboratories, and computational resources.

View large image

“We’re only beginning to understand how artificial intelligence might influence many fields, and this program offers a creative approach for collaboration across many of them,” says Mark Trodden, associate dean of natural sciences at the School of Arts & Sciences. “It’s the type of ingenuity that keeps the School of Arts & Sciences at the forefront of understanding a priority area like AI.”

The new program expands on the data science fellows program that Jain and DDDI co-director Greg Ridgeway, the Rebecca W. Bushnell Professor of Criminology and department chair, launched for School of Arts & Sciences postdoctoral researchers in 2021. Since then, DDDI has served as a nexus for data scientists across more than 25 departments at Penn Arts & Sciences, driving cross-departmental exchange of tools and techniques.

Over the last few years, this program grew to be a thriving cohort of close to 15 postdocs from across the natural and social sciences. In 2025, it has expanded to include over 20 participants with the addition of postdocs from Penn Engineering and, more recently, from the Perelman School of Medicine, through a new partnership with Marylyn Ritchie, vice dean for AI at Penn Medicine, and Li Shen, professor of informatics in biostatistics and epidemiology. The fellows from Penn Medicine are supported by the Institute for Biomedical Informatics.

View large image

“What’s great is when there’s some reason to bring postdocs together that have become experts in small pieces of AI,” explains Kieran Murphy, who is in the inaugural AI x Science cohort and a postdoctoral researcher in Dani Bassett’s Complex Systems Lab. “There’s this knowledge sharing: an awareness of what can be done that just sort of speeds up the spreading of that knowledge.”

Colin Twomey, the executive director of DDDI and a former postdoctoral researcher in Penn Arts & Science who oversees the AI x Science fellowship, notes that these regular gatherings serve as a crucial mixing point. “The postdocs get together on a Tuesday at noon to have lunch, but not only to have lunch,” he says. “There is either a faculty member from different areas of science or engineering that gets invited, or the other format is that postdocs themselves speak about what they do and have a conversation with a technical audience outside their field. We’re building a cycle where novel applications spur new AI developments, and those AI developments enable even broader applications. It creates an environment unlike traditional academic research.”

“The typical postdoc experience is that you are recruited by a faculty member and you work with that faculty member for two or three years in preparation for the next leg of your career journey,” explains Jain, the Walter H. and Leonore C. Annenberg Professor in the Natural Sciences in the Department of Physics and Astronomy at Penn Arts & Sciences. “But there isn’t the kind of experience in which there’s a group of postdocs that just hang out together, explore ideas, and meet weekly to explore possibilities for collaboration outside their thin slice of the universe. Postdocs can be kind of isolated, and we want to change that.”

Playing to Penn’s strengths

Rather than going toe to toe with industry giants on their terms, the program seeks to leverage the University’s unique strengths. “Should we in academia compete for the latest and greatest with industry, or should we be ahead of the curve and try to look for areas that the University is going to have a huge impact, but industry will not necessarily care yet?” asks Vidal, the Rachleff University Professor with joint appointments in the Perelman School of Medicine and Penn Engineering.

View large image

The answer lies in focusing on fundamental questions that may not have immediate commercial applications. “If you’re interested in questions of trustworthy AI—if you’re interested in questions of the social impact of AI developments—then no company currently has the depth of sociologists and psychologists and students of business, along with the AI experts,” adds Jain.

“My personal view is that academics need to wrestle back some of the control—not literal control—but by shaping the narrative around AI and how it can help build a better future for humanity,” says Jain.

Much of the work on AI from the 1950s to around the early 2010s was in academia, but in the last few years, it’s really industry that has been driving things, explains Vidal. “In most industrial applications of AI, the idea of using AI just for the sake of discovery is not necessarily a business proposition … that’s what makes a difference between academia and industry,” notes Vidal. He says that companies, by necessity, focus on research that can lead to products or profits in the foreseeable future, while universities can cast a wider net.

Jain and Vidal acknowledge that some of the best AI minds have been scooped up by industry in recent years, but they’re betting that the intellectual riches of the new program will outweigh some aspects of the tech sector research and prepare the postdocs for a rapidly changing future.

Looking ahead

The early success of the AI x Science fellowship has already sparked plans for growth. This year, the program has grown to include postdoctoral researchers from the Perelman School of Medicine, with Wharton AI fellows expected to join in the fall. Eric Bradlow, vice dean of AI and analytics at Wharton, has been an enthusiastic partner in enabling this next step. Jain, Vidal, Ritchie, and Bradlow are part of Penn’s AI Council, which was assembled by Vice Provost for Research Dawn Bonnell and aims to develop campus-wide initiatives for AI.  

“It’s a natural next step for a university that thrives on collaboration,” Jain says. “The fellowship started by connecting AI and the sciences within Penn, and the vision is to keep extending that network across more fields. Our aim is to reach all the schools and launch the ‘Penn AI fellowship.’”

This would create a “data science hub” where postdocs can dedicate a portion of their time to helping researchers across the university apply AI techniques to their data. “I am a biologist. I have a ton of data, but I don't know machine learning or AI,” Vidal explains as a hypothetical scenario. “If the data science hub has some postdocs, they are doing their main research, but they can dedicate maybe just 20% of their time to these projects.”

And beyond that, Jain and Vidal seek to soon recruit postdocs with expertise in AI for the explicit purpose of bridging collaborations. They acknowledge that this will demand significant resources but remain proud of the program’s achievements to date and are optimistic about its potential to keep growing.