Abstract:

Modern AI faces (at least!) two challenges: (1) building agents capable of autonomy and life-long learning, and (2) embodying them to perform these tasks in the real-world. In this talk, I will discuss our approach to these questions, and show that they also are tightly intertwined with reverse-engineering brains across multiple species, from rodents to non-human and human primates. In other words, by setting the general capabilities of humans and animals as concrete engineering targets, we show that building more capable autonomous agents both advances AI and deepens our computational understanding of large-scale neural populations being collected today (on the order of hundreds of thousands of neurons), thereby forming a tight two-way loop between neuroscience and AI.

Bio:

Aran Nayebi is an Assistant Professor in the Machine Learning Department at CMU, along with appointments in the Neuroscience & Robotics Institutes. His lab, the NeuroAgents lab, works at the intersection of neuroscience & AI to reverse-engineer animal embodied intelligence and build the next generation of autonomous agents, responsibly and safely. Prior to joining CMU, he was a postdoctoral fellow at MIT, and did his PhD in Neurosciences from Stanford with Dan Yamins and Surya Ganguli.