OpenAI CEO Sam Altman and British physicist David Deutsch have agreed on a striking new test for artificial intelligence (AI) – if an AI model could solve the mystery of quantum gravity — and explain its reasoning — it could be considered human-level smart. According to a Business Insider report, the exchange happened on Wednesday (September 24) in Berlin, where Altman visited Axel Springer’s headquarters to meet tech leaders and collect an award. During a fireside chat, publisher Mathias Döpfner asked Sam Altman about his favorite book — Deutsch’s The Beginning of Infinity — and then surprised him by patching the author in live from Oxford.Deutsch, often called the father of quantum computing, has long argued that brute-force training alone cannot produce true minds. He said real intelligence means creating knowledge, not just repeating or remixing it. “Large language models can talk endlessly because they are trained on huge bodies of knowledge,” he explained. “In my mind, genuine intelligence is the ability to create knowledge — spot a problem, invent a solution, test it, and improve it as humans do.”
A new test for AGI
Sam Altman, smiling as Deutsch appeared on screen, asked a direct question: what would it take to convince him that an AI had reached real intelligence? Deutsch pointed to Einstein’s theory of relativity as an example of knowledge creation. “We know he created it because we know his story, what problems he was addressing, and why,” he said.Then Sam Altman offered a hypothetical: “If a future model figured out quantum gravity and could tell you its story — the problems it chose, the reasons it pursued them — would that be enough to convince you?”“I think it would, yes,” Deutsch replied.Altman agreed: “I agree to that as the test.”