Kanaka Rajan

We argue that lethal autonomous weapon systems that use AI are a unique class of weapon system with unique risks that needs novel regulation informed by the AI community in order to mitigate those risks.

We propose methods for studying behavior and neural dynamics in deep RL agents based on neuroscience methods and demonstrate them in a naturalistic simulated foraging environment

We derive well-known learning rules from an objective that casts learning rules as policies for navigating uncertain loss landscapes