LLMs and Mistrust, Discrimination, and Bugs in Software - Yuriy Brun (University of Massachusetts Amherst, USA)

Title: LLMs and Mistrust, Discrimination, and Bugs in Software

Speaker: Yuriy Brun (Professor at the Manning College of Information & Computer Sciences, University of Massachusetts Amherst, USA, and Visiting Researcher at Archimedes Research Unit, Athena Research Center, Greece)

Abstract: Software is ubiquitous, and trusting it is no longer optional.

Unfortunately, modern software has been caught lying, discriminating, and even causing deaths. This talk will explore how we can measure trust in software, understand what factors affect trust, and increase software trustworthiness. First, I'll describe how trust games, an instrument from psychology, can identify what affects users' trust in software. Second, I'll introduce Seldonian algorithms that fundamentally re-envision machine learning to produce models that are probabilistically guaranteed to satisfy fairness and safety requirements, even on unseen data. Third, I'll show how cutting-edge natural language processing, including LLMs, hold the key to automatically proving software correctness, the ultimate goal in software trustworthiness. Overall, I'll describe significant progress toward understanding software trust and improving software trustworthiness and lay out the challenges that lay ahead.

Short Bio: Yuriy Brun is a Computer Science Professor at the University of Massachusetts Amherst and a Visiting Researcher at Archimedes Research Unit, Athena Research Center, Greece. He is a Fellow of the IEEE and a Distinguished Member of the ACM. His research focuses on trust in software systems, including how improving software quality and fairness affect trust, and how formal verification can increase trust. He received his PhD from the University of Southern California in 2008 and was then a Computing Innovation postdoctoral fellow at the University of Washington. Prof. Brun is a recipient of the NSF CAREER Award, the FSE Test of Time Honorable Mention Award, the SEAMS Most Influential Paper Award, the IEEE Computer Society TCSE New Directions Award, and the IEEE TCSC Young Achiever in Scalable Computing Award. Eight of his papers have won Best Paper, Distinguished Paper, or Distinguished Artifact awards.

He has received a Google Inclusion Research Award and a Google Faculty Research Award, an Amazon Research Award, a Microsoft Research Software Engineering Innovation Foundation Award, a Lilly Fellowship for Teaching Excellence, and a College Outstanding Teacher Award. He founded the Rising Stars in Computer Science program at UMass Amherst (https://www.cics.umass.edu/rising-stars) and his recent Science paper (http://doi.org/10.1126/science.aag3311) introduced machine learning methods that guarantee safety and fairness. https://people.cs.umass.edu/~brun/

________________________________________________________________________________

________________________________________________________________________________