Del Toro Says AI Can Reach the Middle Fast — But the Final 10% Is Pure Human Alchemy

“They are alchemy.” That is how Guillermo del Toro described the moments of creative accident — the unexpected, non-linear collisions of influence and instinct — that give great art its singular quality. And it is, he argues, precisely what no AI can replicate. The Pan’s Labyrinth director made the remarks in response to a fan question asking what the “biggest thing lost” is when a computer generates an image instead of a human hand. His answer has circulated widely since it was shared online this week.

The Pale Man and the Manta Ray

Del Toro did not reach for abstractions. He used one of cinema’s most recognisable creature designs to illustrate his point: the Pale Man, the eyeless, hand-faced monster from Pan’s Labyrinth. The Pale Man was originally sculpted as a conventional old man with a face. The character transformed when, at two separate moments spread across a week or a month, Del Toro encountered a photograph of a manta ray and the poster for Don Coscarelli’s Phantasm. Neither image had anything to do with the creature he was building. Together, they changed everything.

“That combination, which happened at two different times during a week or a month, are not permutations that are chemically part of an equation. They are alchemy. And then I said, ‘Take out the face.’ And then I said, ‘Let’s put the eyes on the hands.'”

This triangulation happened because Del Toro spent a month or two months living inside the problem. It was not linear. It could not have been prompted. The creative outcome existed precisely because of the time spent in uncertainty — and the accidental collisions that uncertainty made possible.

What AI Cannot Access

Del Toro’s critique is precise rather than reactive. He frames AI as “a super-fast engine that makes semi-sentient permutations of things that are already done.” Creation, he argues, is categorically different: things that have never existed, coming into being for the first time. The second loss he identifies is apprenticeship. “You’re never a master; you’re always an apprentice,” he said, citing this as a kinship he shares with Christopher Nolan. The slow accumulation of craft — through years of working on difficult problems with human hands and human judgment — cannot be shortcut without cost. Cut a generation of artists off from learning their trade, and the knowledge accumulated across a medium’s entire history starts to thin. A third element is the accident. “You’re gonna find that 30% of a great image occurs accidentally,” he said. “A drop of water hits the lens, or the camera shakes.” These things cannot be prompted — they occur from the interaction with reality, with real material, with physical presence in a moment.

A Year of Speaking Out on AI

This is not Del Toro’s first statement on AI in 2026 — it is closer to his fifth or sixth. At San Diego Comic-Con in July, he revealed that the 3D re-release of Pan’s Labyrinth had been produced through a manual rotoscoping process involving roughly 1,000 artists. “Obviously I said, ‘Absolutely no goddamn AI,'” he told the packed Hall H audience. At Cannes in May, presenting the 4K restoration of the same film, he was more polemical. He drew a direct line between Pan’s Labyrinth’s resistance themes — a child defying a brutal system through imagination — and what he called the cultural threat posed by AI replacing human creative workers. He closed the speech with “Never, never, never give to fear.” Earlier in January, he told NPR he would “rather die” than use generative AI in his work. What distinguishes Del Toro’s position from many contemporaries’ is its specificity. He is not primarily arguing about jobs, or legal rights, or contracts — though those arguments have their place. He is arguing about what art is, what making it does to a person, and what is quietly destroyed when the process is compressed into a prompt.