Robert Wolfe’s research develops technical and sociotechnical approaches to improving the reliability and transparency of modern general-purpose AI models. His work creates novel quantitative methods to investigate and mitigate vulnerabilities in AI, as well as approaches that foreground small, open models as more transparent alternatives to proprietary systems. Wolfe adopts qualitative and mixed methods to understand the uses, shortcomings, and opportunities of the general-purpose models in information work, contextualizing his quantitative work in the experiences of real-world organizations and communities.
Wolfe’s research has been published at prestigious venues including the ACM Conference on Fairness, Accountability, and Transparency; AI, Ethics, and Society; and AAAI. Wolfe has given invited talks on his research at venues including the National Institute of Standards and Technology and the Santa Fe Institute for Complex Systems. At the University of Washington, he was a founding member of The AI Clinic, an organization dedicated to supporting academics and community members in the responsible use of general-purpose AI. Wolfe’s graduate work was supported by grants from organizations including the UW iSchool’s Strategic Research Initiative and Google Research.
Wolfe’s ongoing research studies evolving privacy norms around chatbots, applications of AI in high-stakes information work, and uses of AI on social media. His recent collaborations on topics including privacy attacks on large language models, calibration methods for general-purpose AI, justice-oriented neighborhood technologies, trust-building approaches on social media, and novel approaches to AI literacy have been published in venues including ACM CHI, CSCW, SOUPS, ICML, ACL, and IDC.