Demis Hassabis thinks AGI is coming by 2030 — and he's already thinking about what comes after.
Speaking at Stanford's Graduate School of Business, the Google DeepMind CEO put a date on what many consider the most consequential moment in human history: artificial general intelligence, the point where machines can match human cognition across the board. "I believe we're only a few years away — maybe 2030, plus or minus a year," he said. "It's going to effectively be a new human era."
He didn't stop there. A decade out, Hassabis believes we'll look back and realize we were standing at the base of something much larger — the technological singularity, where AI doesn't just match human intelligence but surpasses it and begins improving itself beyond our control.
Not everyone agrees on the timeline. Sam Altman has written about humanity being close to "digital superintelligence." Anthropic's Dario Amodei put powerful AI as early as late 2026. Shane Legg, Hassabis's own chief AGI scientist at DeepMind, gives 2028 a 50% probability for even "minimal AGI." And on the skeptical end, Meta's Yann LeCun has called the entire concept of general intelligence "complete BS," arguing today's large language models simply won't get there.
But Hassabis is less interested in the debate than in the preparation. "Society needs to hear that we don't have long to get ready," he said. "And it's going to be enormously profound."
So what happens to people? In a separate interview, Hassabis pointed to the qualities that machines are unlikely to replicate anytime soon: taste, design sensibility, original thinking, and the ability to synthesize ideas across disciplines. Those who cultivate these skills, he said, will be in a strong position.
His underlying optimism is almost philosophical. Humans, he points out, are already general intelligences — ones who built civilization with brains wired for hunting and gathering. "Why would we stop here?" he asked. The arrival of AGI, in his view, isn't an ending. It's an accelerant.