Geoffrey Hinton warns AI is “dangerous” in Nick Holt’s Tribeca documentary

Nick Holt’s Tribeca documentary “AI: Probably Nothing to Worry About” opens with a deceptively calm figure: Geoffrey Hinton, a 78-year-old Nobel Prize winner who is introduced as a mild-mannered man who spends his days “communing with nature.” The charm ends fast. Holt then makes the premise of the film impossible to ignore, because Hinton, described as the architect of deep learning, says his “main mission now is to warn people how dangerous” an “apex intelligence” could be.

Hinton does not couch it in marketing language. He compares the AI moment to the atomic age, musing, “Maybe the wisest decision in 1946, would have been don't develop H-bombs.” The documentary’s subtitle is the biggest lie in the room. Holt’s point is blunt: the most unsettling thing in AI may not be the algorithm itself, but the humans who created it, changed their minds about it, and then kept pushing it into the world.

From there, the film lays out what it calls a kind of “murderer's row” of AI’s founding fathers, grounding the tension in personal reversals and shifting beliefs. Holt moves from Hinton to Demis Hassabis, introduced via old footage as an adorable, chess-playing 9-year-old, then as a young entrepreneur thrilled when he sees his program play and win Pong. Today, Hassabis is the CEO of Google DeepMind, and the documentary includes him considering computers “almost a magical extension” of our brains. The narrative contrast matters. The early wonder remains, but the emotional temperature changes as the people who built the tech start raising alarms.

Next comes Elon Musk, our familiar “who-whoops-started out calling AI our biggest existential threat.” Holt also includes Musk’s later framing that, “with artificial intelligence, we are literally summoning the demon.” The documentary then pivots to OpenAI co-founder Sam Altman, showing him telling a congressional panel: “My worst fears are that we cause significant harm to the world. If this technology goes wrong, it can go quite wrong.” In other words, the film is not only about invention. It is about accountability under institutional stress, and what happens when the person who understands the power is no longer confident the rest of society can manage it.

To make the stakes feel concrete, Holt uses a reference point that every executive in AI governance recognizes immediately, even if they disagree about conclusions: a clip of “2001: A Space Odyssey” and HAL. The documentary lands on Hinton’s question: “How many examples do you know of more intelligent things being controlled by less intelligent things?” That question is the film’s engine. It frames AI risk as a mismatch between capability and oversight. It implies a systems-level failure mode: the more the technology improves, the more the governance around it can lag behind, even when regulators and boards are trying.

The documentary also reminds viewers that the incentives around AI are still messy and politically charged. It notes that Sam Bankman-Fried, described as one of the earliest investors in Anthropic, which brought us Claude, is currently in jail. It also mentions that Musk recently and very publicly sued Altman and OpenAI, and “lost.” Whether you read that as justice, business conflict, or both, it underlines the film’s broader anxiety: when AI is moving this fast, disputes, legal battles, and reputational storms are not side plots. They become part of the governance landscape that leaders must navigate.

Holt’s execution is what makes the whole thing stick. He spends almost no time on potential positives of AI, referring only glancingly, for example, to medical benefits. He also does not try to be a catalogue of pros and cons or a neutral explainer. Instead, “AI: Probably Nothing to Worry About” behaves like “a vibrating entity of freeform anxiety,” a deliberate style choice that Holt uses to build urgency. Near the end, the documentary escalates its framing with a question that effectively forces the viewer to stop treating the subtitle as a joke: “Geoffrey, do you think we're all fucked?” Hinton responds, “Well, it depends,” and continues, “We might be. I really don't know. When I'm feeling slightly depressed, I think people are toast.”

If you’re an executive, the strategic lesson is not that Hinton is right or wrong. The lesson is that the people closest to the technology are publicly describing existential-scale risk while regulators, boards, and enterprise buyers still often run on slower cycles. That gap is where decision-makers live. Deals get signed while models get smarter. Policies get drafted while competitors ship. And when the creators themselves describe the worst case, the boardroom has to answer a hard question: what is your oversight plan when the next performance jump arrives before your governance does?

Holt, a quadruple BAFTA nominee, insists on that question by refusing to soften it. The film closes with the feeling that the future is not abstract. It is in the hands of humans who built the systems, then began warning about them, and now have to live with what they unleashed. For leaders making AI decisions today, the documentary is essentially an ultimatum disguised as entertainment: don’t treat AI risk as a distant scenario. Treat it as an active governance problem right now.