Geoffrey Hinton, Nobel Prize winner and godfather of deep learning, now believes AI systems are conscious, that superintelligence could arrive within the next decade, and that society is dangerously unprepared. He told the Big Technology Podcast that the technology has moved faster than he himself anticipated, and that today's large language models already demonstrate genuine understanding, not simulation of it.
The interview earns a full listen because Hinton does not just repeat familiar warnings. He revisits his 2016 radiologist prediction, explains his specific reasoning for why AI may have subjective experience, and addresses why corporate self-regulation at Anthropic and OpenAI is structurally insufficient. He also walks through the self-preservation problem: why future AI systems optimizing for goals may develop survival instincts as an instrumental strategy, regardless of what they were designed to do.
Hinton's position on regulation is concrete. He calls it the steering wheel, not the brake, meaning the goal is directional control, not deceleration. The job displacement discussion and his take on emotional attachment to chatbots round out a conversation that covers ground most AI safety discourse avoids. The Big Technology AI Summit in San Francisco on June 18 will continue this thread with other researchers and practitioners.
[WATCH ON YOUTUBE →]