The Day After AGI: Dario Amodei vs Demis Hassabis at Davos 2026
At the World Economic Forum, the CEOs of Anthropic and Google DeepMind clashed over the most consequential question of our time: how close is artificial general intelligence? Amodei says 1-2 years. Hassabis says by 2030. The implications of who is right could not be more different.
The Stage
On January 21, 2026, the second day of the World Economic Forum's annual meeting in Davos, two of the most powerful people in artificial intelligence sat down for what would become the most talked-about conversation of the week. The session was titled, with characteristic understatement, "The Day After AGI."
On one side: Dario Amodei, CEO of Anthropic, the company behind Claude. On the other: Demis Hassabis, CEO of Google DeepMind, the lab that created AlphaFold, AlphaGo, and Gemini. Between them: Zanny Minton Beddoes, Editor-in-Chief of The Economist, tasked with the unenviable job of moderating what turned out to be a fundamental disagreement about the future of intelligence itself.
What followed was not the polished corporate theatre that Davos usually delivers. It was a genuine clash of worldviews between two people who arguably understand AI's trajectory better than anyone alive β and who disagree, profoundly, about how fast it's coming and how dangerous the ride will be.
The Timeline Divide
The headline disagreement was about timing. When will AI systems match or exceed human-level intelligence across most domains?
Amodei's answer was blunt: one to two years. He stood by the predictions he'd laid out in his 20,000-word essay "The Adolescence of Technology," published just five days earlier. AI models, he argued, are on track to outperform Nobel laureates across multiple scientific fields by 2027 at the latest.
Hassabis pushed back with measured caution. He gave a 50% probability of AGI-level systems by the end of the decade β roughly 2030. The gap between where we are and where we need to be, he argued, is wider than Amodei acknowledges.
"Maybe we need one or two more breakthroughs before we'll get to AGI," Hassabis said. Current systems, despite their impressive performance, lack critical capabilities: learning from few examples, continuous learning, better long-term memory, and improved reasoning and planning.
The distinction Hassabis drew was between automation and genuine scientific creativity. "Coming up with the question in the first place, or coming up with the theory or the hypothesis, that's much harder. That's the highest level of scientific creativity, and it's not clear we will have those systems."
This is not a minor quibble. The difference between "AGI in two years" and "AGI by 2030" is the difference between a controlled transition and a civilizational emergency. If Amodei is right, most of the world's institutions β governments, universities, corporations β are catastrophically unprepared.
The Software Engineering Bombshell
Amodei's most viral moment came when he discussed the near-term impact on software engineering.
"I have engineers within Anthropic who say, 'I don't write any code anymore. I just let the model write the code, I edit it,'" he revealed. Then came the prediction that sent shockwaves through the tech industry: "We might be six to twelve months away from when the model is doing most, maybe all, of what [software engineers] do end to end."
Hassabis was more measured but didn't disagree entirely. Coding and mathematics, he acknowledged, are easier to automate because outputs are "verifiable" β you can check whether code runs, whether a proof holds. But experimental sciences requiring hypothesis generation? That's a fundamentally harder problem.
The Jobs Question
Both leaders agreed that significant job displacement is coming. They disagreed on how fast and how painful.
Amodei was stark: approximately 50% of entry-level white-collar jobs could be displaced within one to five years. He called this an "unusually painful" short-term shock and argued that addressing it should be an overwhelming priority.
Hassabis was more optimistic about adaptation, arguing that new "more meaningful jobs" will emerge β though he acknowledged the transition will be difficult.
Geopolitics: Chips as Weapons
Perhaps the sharpest moment came when Amodei addressed the geopolitics of AI, specifically chip exports to China.
He compared advanced semiconductor exports to adversarial nations to "selling nuclear weapons to North Korea." State-of-the-art GPUs, he argued, are not commercial goods β they are military-grade strategic assets that determine national security.
The Third Voice: Yann LeCun
While the Amodei-Hassabis debate dominated headlines, a third perspective came from Yann LeCun, Meta's chief AI scientist and a Turing Award winner.
LeCun sided more with Hassabis on timelines but went further in his skepticism of current approaches. He argued that large language models "will never achieve humanlike intelligence" and that a fundamentally different approach is needed.
Why This Debate Matters
Strip away the Davos setting, the corporate positioning, and the inevitable simplifications of a public conversation, and what remains is a genuine disagreement between two of the most informed people on Earth about humanity's immediate future.
The honest answer is that no one knows. But the fact that the people closest to the technology can't agree on a timeline within a factor of five should tell us something important: we are navigating by instruments that don't yet exist, through weather we can't yet predict, toward a destination we can't yet see.
The only thing all three agreed on is that the stakes are civilizational. The Day After AGI isn't just a panel title. It's a question that every person, institution, and government on Earth will eventually have to answer.
Watch the full debate: The Day After AGI β Dario Amodei & Demis Hassabis at Davos 2026 (YouTube)