ISP AI Forum: Nihal Murali and Dr. Ahmad Tafti

Join us for the ISP Forum on Friday, September 25, 2026, in SENSQ 5317! Lunch and refreshments start at noon, with the forum running 12:30–1:30 p.m. We will be featuring Nihal Murali and Dr. Ahmad Tafti. See presentation details below. Nihal Murali’s Presentation Information Title: When Should Humans and AI Work Together? Designing Hybrid Systems for Clinical Decision-Making Abstract: AI is increasingly capable of supporting clinical decision-making, but an important question remains: how should humans and AI work together when neither is perfect and expert attention is limited? This talk explores this question in the setting of detecting cerebral ischemia (reduced blood flow to the brain) during carotid artery surgery. This is a high-stakes monitoring task in which changes in brain electrical activity must be recognized quickly, yet trained neurophysiologists who perform this monitoring are limited in number. We develop a hybrid novice–AI system that combines an AI model with assessments from minimally trained novices with no prior experience in interpreting brain-wave recordings. Despite this limited training, the hybrid system learns when to rely more on the novice and when to rely more on the AI model, allowing the two to complement each other. Remarkably, the combined system not only improved substantially over the AI model alone, but also achieved performance statistically non-inferior to expert interpreters. Our study points toward a practical vision for clinical AI: not replacing clinicians, and not requiring constant human supervision, but designing systems that learn when human judgment adds value and how best to combine it with AI. Bio:Nihal Murali is a 6th year Ph.D. candidate in the Intelligent Systems Program at the University of Pittsburgh, jointly advised by Prof. Shyam Visweswaran and Prof. Kayhan Batmanghelich. His research lies at the intersection of machine learning, medicine, and human–AI collaboration, with the broader goal of developing