Artificial intelligence is reshaping industries at a breakneck pace, transforming how companies inform decisions and unlock fresh opportunities. That momentum is also redefining research possibilities in academia. At Wharton, doctoral students are charting new territory, pursuing lines of inquiry that would have been difficult to imagine just a few years ago.
A third-year PhD student in Wharton’s marketing department, Roger Saumure GRW28, examines how AI shapes communication and behavior. In one project, conducted with Robert Meyer, faculty co-director of Wharton Human-AI Research and one of Saumure’s advisors, Saumure investigated what happens when people ask these tools to revise their writing. In the experiment, participants composed online reviews and then viewed AI-generated revisions. “Consumers are very willing to adopt these writing suggestions, which can distort what they want to say,” Saumure says. “We need to be aware of how large language models are changing sentiment.” The findings suggest that users who begin with negative views are more likely to accept edits than users who start with positive ones.
In a different study, with associate marketing professor Marissa Sharif, Saumure is exploring whether AI can influence real-world action. Partnering with university gyms, he is testing if checking in with an AI coach about exercise goals can increase attendance, compared to checking in with a person or no check-in at all. “If the AI does just as well as a human,” he says, “that would point to a scalable solution to help people change behavior.” Both of Saumure’s projects have received funding from the Wharton AI & Analytics Initiative’s AI Research Fund and the Mack Institute for Innovation Management.
Also in marketing, Weixin He GRW28 is using AI to rethink product design and personalization. Supported in part by the AI Research Fund, she and assistant marketing professor Ryan Dew are developing methods that use short, adaptive, personalized surveys to estimate a consumer’s tastes and generate tailored product concepts. “Designing good products requires understanding consumer preferences across various product attributes,” He says. In categories like fashion, where features such as sizing, patterns, and style can vary widely, her framework aims to identify what a specific customer likes and produce customized designs in real time. She is also working on a related effort to personalize content such as news headlines, balancing engagement with editorial constraints.
In the operations, information, and decisions department, Ruben R. Salas GRW28 is studying human-AI collaboration, including how AI is altering creative processes and how we experience digital markets. In one line of research, Salas and advisor Kartik Hosanagar, faculty co-director of Wharton Human-AI Research, examined how large language models perform at generating humor. Their paper — which received the Best Paper Award at the INFORMS Conference on Information Systems and Technology last year — explored whether AI could help people craft witty captions for New Yorker cartoons. “We found that AI increases the quantity of ideas you produce and can improve their quality during refinement,” Salas says. “But collectively, we’re decreasing the diversity of those ideas.”
His newer research — which, like his first project, is supported by the AI Research Fund — focuses on what is known as “generative engine optimization.” As consumers shift from clicking search links to asking AI systems for recommendations, firms are beginning to tailor product descriptions not just for human readers, but for the models that deliver answers. “You’re not only writing for a human anymore,” Salas says. “You’re writing to influence the model.” For this analysis, Salas and Hosanagar have drawn on Internal Wharton Research Data Services, a catalog of datasets made available to Wharton and Penn faculty and students for academic purposes that is also known as iWRDS. “That’s one of the reasons I chose Wharton,” Salas says. “You have access to so many data resources to build this kind of research.”
Eric Bradlow, vice dean of AI and analytics at Wharton and He’s advisor, sees the ecosystem of support that Wharton has built around AI research as one of the School’s greatest strengths. “Through initiatives like the AI Research Fund, access to robust data resources such as iWRDS, and deep cross-disciplinary collaboration, our doctoral students have the infrastructure to pursue research that is not just ambitious but genuinely consequential,” he says.
Applying AI is only part of the story. In the statistics and data science department, Yu Huang GRW28 is working to understand how the technology itself reasons through complex tasks. In one paper, she and her co-authors show that the structure of a problem plays a decisive role in whether the model can extend reasoning beyond its training. “The main takeaway,” she says, “is that the task structure determines how long the reasoning can generalize.” In more recent work, Huang has studied reinforcement learning with verifiable rewards, or RLVR, a method that provides feedback only at the end of a task, much like rewarding a dog after it carries out a command. She explores why RLVR is most effective at the “edge of competence,” when problems are just difficult enough to stretch a model’s reasoning ability.
Nancy Zhang, vice dean of Wharton doctoral programs, says this moment of AI advancement is an inflection point for doctoral study that is leading to new questions and ways of doing research: “At Wharton, our PhD program encourages intellectual risk-taking and cross-disciplinary exploration, giving students the flexibility, mentorship, and technical depth needed to define emerging fields rather than simply follow them.”
Huang’s trajectory reflects that philosophy. She arrived at Wharton in 2023, just as generative AI entered the mainstream, initially concentrating on theoretical machine learning. But as large language models reshaped the broader AI landscape, her agenda changed, too. She describes Wharton as a place where faculty were already thinking ahead about LLMs and encouraging new directions. “I wanted to understand what my theory could contribute in this area,” Huang says.
Salas says that same culture has shaped his own experience: “It’s the full support — from the resources to develop the research to the funds that help strengthen its impact through conferences and other chances to present.”
Published as “The AI Doctors Are In” in the Spring/Summer 2026 issue of Wharton Magazine.
