There’s A 50% Chance That Einstein-Level AI Systems Will Be Created In 5 Years: Google’s Demis Hassabis

Many AI experts are predicting that AGI is just a couple of years away, but Google DeepMind CEO Demis Hassabis seems to be more conservative in his outlook.

Google DeepMind’s Demis Hassabis has said that there’s a 50 percent chance that AI systems as smart as Einstein will be created in the next five years. “The way I would define AGI is a system that can exhibit all the cognitive capabilities humans can,” he said at an event. “And that’s important because the human mind is the only example maybe that we know of in the universe that is a general intelligence,” he added.

“Of course, how to test that is the big question. The one I’m really looking for — and why I think it’s a little bit further out, maybe 50 percent chance of in, in five years time, so maybe by the end of the decade — I think we don’t have systems yet that could have invented general relativity when Einstein did with the information that he had available at the time. Or another example I give is, can you invent a game like Go? Not just play a great move like move 37, or build AlphaGo that can beat the world champion, but could you actually invent a game that’s as beautiful aesthetically as Go is. So I think it’s going to take a little bit longer to get that kind of capability,” Hassabis said.

Hassabis seemed to be saying that AGI shouldn’t only be able to complete all tasks that humans can, but also lead to breakthroughs that exceptional human beings, like Albert Einstein, were capable of. Hassabis said that current AI systems were quite powerful, but couldn’t come up with novel ideas, like how Einstein had introduced general relativity in 1915. Also, AI systems were currently good at playing games like Go, but hadn’t yet displayed the creativity to create a new game like Go. Hassabis believes that an AGI should be able to do these things, and feels that it could take longer than many experts are currently predicting. Some people have very aggressive timelines for AGI — Anthropic CEO Dario Amodei has said that AGI will be achieved in 2026-27, and Softbank CEO Masayoshi Son has said it’ll be achieved even sooner. But Demis Hassabis has said that there’s only a 50 percent chance that AGI will be achieved by 2030, indicating that superhuman intelligence might still be some way away.

Posted in AI