Robert Wolfe
I’m an Assistant Professor at Rutgers University in the School of Communication and Information (SC&I), where I study the epistemic risks and opportunities of modern AI systems. My research considers:
- Small-scale approaches to generative AI that can help researchers and practitioners avoid dependence on proprietary technologies, especially in situations where sensitive data is involved, or reproducibility is paramount.
- Creating novel dimensions in the design space of generative technologies to help maximize transparency in high-stakes epistemic settings such as fact-checking.
- Understanding the nature of undesirable AI biases related to protected demographic characteristics such as race and ethnicity, nationality, age, and gender, most often in multimodal language-and-image models.
news
| Aug 26, 2025 | Excited to share that two new papers probing social norms surrounding human interactions with AI chatbots have been accepted to AIES 2025. The first is a study drawing on the principles of Nonviolent Communication to inform AI-mediated interactions, and the second is a study of privacy norms among U.S. users of LLM-based chatbots. Congrats to the amazing group of students whose work in the User Empowerment directed research group made the second study possible, and especially to the talented first author Sarah Tran! |
|---|---|
| Aug 17, 2025 | Glad to have been a part of two creative new NLP papers probing the limits of LLMs! The first is a new paper from Bin Han studying the capabilities of LLMs to perform spatial integration tasks on urban data, and will appear at COLM 2025. The second was led by Dexter Xu and Bingbing Wen and presented at ACL Findings 2025; this study develops new approaches to confidence estimation in LLMs, and examines whether LLMs mirror human patterns of overconfidence. |
| May 22, 2025 | Two new papers collaborating with the visionary social computing researcher JaeWon Kim! Check out her work on reducing dysfunctional fear among adolescents on social media at CSCW 2025, and her paper introducing the trust-enabled privacy design framework at SOUPS 2025. |
| May 1, 2025 | Excited to share that our research developing new methods for understanding privacy threats in large language models has been accepted to ICML 2025. Congrats to two fantastic lead authors, AI privacy and fairness researcher Lucas Rosenblatt and deep learning and spatial computing researcher Bin Han! |
| Apr 25, 2025 | Proud to have been a part of research on reconceptualizing neighborhood safety technologies to meet the needs of communities with the amazing social justice and community technologies researcher Ishita Chordia at CHI 2025 - where it also received a Best Paper Honorable Mention! |