The holy grail of technology — artificial general intelligence (AGI) that can match or outdo humans — is on the horizon, Google DeepMind co-founder and CEO Demis Hassabis says.
- But the risks of something going seriously wrong are also in sight, and some are even happening now, he warns.
The big picture: Google set the entire AI world spinning in recent months with the giant leaps in its frontier model Gemini, prompting a "code red" at OpenAI and forcing others to rethink the competitive landscape.
- But Hassabis,49, who won the Nobel Prize in Chemistry last year, and his DeepMind team are already thinking far, far ahead.
"We're definitely not there now" in terms of AGI, Hassabis said in an interview with Axios' Mike Allen at the Axios AI+ SF Summit in San Francisco on Thursday.
- "Quite close. I think we're like 5 to 10 years away if you were to ask me," he said.
Between the lines: The Nobel-winning Hassabis said the current AI technology — large language models like Gemini — needs to scale up, at a minimum, to get to an AGI future.
- But even that may not be enough.
- "My guess is, if I was to guess from my vantage point now, is that one or two more big breakthroughs — there's innovation going on all the time, by the way, even including in scaling our existing techniques — but I'm talking like a Transformer level or AlphaGo level type of breakthrough," Hassabis said, referring to two of Google's key historic technology achievements.
- "I suspect when we look back once AGI is done that one or two of those things were still required, in addition to scaling."
Yes, but: The road to AGI will be littered with missteps, including bad actors. Some "catastrophic outcomes" like cyberattacks on energy and water infrastructure are a clear and present danger.
- "That's probably almost already happening now, I would say, maybe not with very sophisticated AI yet, but I think that's the most obvious vulnerable vector," he said.
- That's one reason, the London native added, that Google is so heavily focused on cybersecurity, to defend against such threats.
Zoom out: Whatever happens between now and AGI, a running theme of the AI+ summit was the inevitability of AI-driven change, and the need to accept it and adapt.
- "There'll be a tremendous amount of AI built and pretty much anybody in this room, or not in this room, our jobs will be done fundamentally differently in five years, including mine," Palo Alto Networks CEO Nikesh Arora said.
- Sierra co-founder Bret Taylor said of the AI boom: "I think it would be a mistake to dismiss it as snake oil." Even as he acknowledged it was right to ask questions about a bubble, he added: "There's going to be a handful of companies that are truly generational."
- Box CEO Aaron Levie said the technology industry benefits from having as many as five viable AI competitors at the same time, fighting what amounts to a daily skirmish for technical supremacy. "If you look at it over a five-year period, it's sort of completely anybody's game," he said.
The bottom line: AGI is coming, and its potential to change humanity is almost limitless.
- "The best-case scenario I've always dreamed about and worked my whole life ... is a kind of radical abundance — this idea we've solved the biggest issues confronting society and humanity today," Hassabis said.
