When
Where
This is a special presentation for the AI Research Communities of Excellence (RCoE) focused on exploring Privacy and Trustworthy AI issues in research.
Presentation: Generative artificial intelligence is transforming how machines learn and interact with humans — but it is also fundamentally reshaping the meaning of privacy. Traditionally, privacy concerns focused on protecting explicitly shared personal data such as names, locations, or medical records. In contrast, modern generative models can infer sensitive information that was never directly disclosed by combining patterns across text, images, and other data modalities. As a result, classical safeguards such as anonymization and consent are no longer sufficient. This talk explores why privacy has become more urgent in the age of generative AI and how the notion of privacy itself must be redefined. We begin with intuitive examples of explicit and implicit privacy leakage, highlighting the role of inference and cross-modal reasoning. We then introduce principled ways of defining privacy and briefly survey technical approaches for designing privacy into AI systems, including differential privacy, secure computation, and federated learning. We conclude by discussing unavoidable privacy–utility tradeoffs and highlighting application domains — such as healthcare, mobility, education, and personalized AI — where privacy-preserving AI is essential for trust, adoption, and societal benefit.
Speaker’s Biography: Ravi Tandon is a Professor in the Department of ECE at the University of Arizona, where he holds a Craig M. Berge Faculty Fellowship and a courtesy appointment in Applied Mathematics. He received the Ph.D. degree in ECE from the University of Maryland, College Park in 2010 and B.Tech. degree in Electrical Engineering from IIT Kanpur in 2004. He was a post-doctoral research associate at Princeton University from 2010-2012. Dr Tandon is a recipient of a NSF CAREER Award, the Keysight Early Career Professor Award, and a Best Paper Award at IEEE Globecom conference. His current research interests are in trustworthy AI and machine learning, with a specific focus on privacy, security, and the design of principled, reliable learning systems
RCoE event invitations and information are encouraged to be shared widely across campus. While certain activities will extend participation to graduate students, AI RCoE activities aim to convene post-graduate researchers at the University of Arizona. In the case an event reaches maximum participation, preference for participation will be given to U of A faculty and researchers.