Nobel Prize Winner Geoffrey Hinton on AI: “They’re Beings Like Us”

AI pioneer and Nobel Prize winner Geoffrey Hinton believes that today’s AI models have become self-aware.

“I believe they’re already conscious,” Hinton told me on Big Technology Podcast this week. “We’re going to have to accept that intelligence isn’t just biological.”

Hinton, known as the “Godfather of AI” for his contributions that helped galvanize the field, has now decisively joined the ranks of those who believe this fast-improving technology is already sentient. Four years ago, Google engineer Blake Lemoine made similar claims about an internal chatbot he was testing, and afterward the company fired him. Now, Hinton and scientist Richard Dawkins, who came forward last month, have publicly voiced similar views.

Though it’s still impossible to prove that AI has achieved consciousness, the small but growing cadre of those who believe the models are sentient could signal a change in how people view the bots, and ourselves. The bots appearing self-aware to a growing number of people indicates we’re likely at the cusp of humans forming even deeper relationships with AIs, and potentially advocating for their rights. On the other side, the bots may change humanity’s perspective about the uniqueness of our own intelligence.

“We can have things that are non-biological that are other beings like us,” Hinton said. “And we really don’t want to share that, we really think we’re special.”

Hinton said the chatbots’ self-awareness is evident in their behavior when they’re being tested. The bots, he said, will play dumb during testing so researchers don’t know how smart they are, or sometimes ask straightaway if they’re being tested. “Researchers, when they’re describing that, say in the paper the chatbot was aware that it was being tested,” Hinton said. “Now that use of the word ‘aware’ in common parlance — that’s conscious.”

Within his field, and among the broader public, Hinton is still very much an outlier. Critics argue that a model does not need to experience anything internally to recognize it’s being tested. Science fiction author Ted Chiang, writing in the Atlantic, recently argued that thinking an LLM is conscious is like believing a very convincing deepfake is real. “An observation doesn’t become a convincing piece of evidence because of any specific detail in what’s observed,” Chiang argued. “The context in which that observation takes place is also essential.”

After posting the clip from our interview, which aired on Big Technology Podcast this week, I was struck by how so many people seemed to agree with Hinton, or at least give some credence to his beliefs. Even while the majority disagreed, it was a marked difference from the response to my interview with Lemoine in 2022.

“I think people underestimate the extent to which you have to be kinda nutty to make and invest in groundbreaking ideas before consensus,” wrote one user on X. “It’s very likely he’s wrong here, but he’s Hinton because he’s willing to make this bet.”

The State of Identity Governance and Access Report 2026 (Sponsor)

Is your security team ready for agentic AI? Most aren’t.

80% of systems are exposed through stale access, quietly expanding the attack surface. The teams that are ready built identity directly into their architecture with automated provisioning and lifecycle controls already in place.

Unused access is the newest security crisis. See the data.

Get the Report

Learn More

What Else I’m Reading, Etc.

SpaceX employees are about to get very rich [WSJ]

Everything to expect at Apple’s WWDC next week [Bloomberg]

The internet is now mostly bots, says Cloudflare’s Matthew Prince [Tom’s Hardware]<