Eric Zelikman

AI researcher Eric Zelikman helped pioneer the reinforcement learning techniques powering today's most advanced models — but he says the field has been solving the wrong problem all along. Instead of building AI that replaces human effort, he shows why we should build technology that can actually understand, collaborate and partner with us. Learn more about the limitations of current systems (like sycophancy and rigid automation) and his proposal for a more human-centric approach.