Tuesday, May 19, 2026, on the stage of the Shoreline Amphitheatre in Mountain View, Demis Hassabis uttered a sentence that still echoes throughout the industry: “When we look back, I think we will realize that we were at the foothills of the singularity.” The founder and CEO of Google DeepMind, a Nobel laureate in Chemistry in 2024 for AlphaFold, immediately caused astonishment in the room. The day after the event, The Verge, Semafor, and Business Insider relayed the statement worldwide – not as a provocation, but as a major rhetorical shift.

Hassabis Breaks Semantic Taboo

For years, Demis Hassabis was one of the few AI leaders to resist transhumanist prophecies. When OpenAI's Sam Altman spoke of imminent AGI, Hassabis remained cautious. When investors fantasized about the singularity, he tempered expectations. But at I/O 2026, he deliberately changed his tone. “We debated this phrasing internally. I decided to be authentic about what I truly believe,” he confided in an interview with Semafor published on May 20. For the head of DeepMind, Artificial General Intelligence – a system capable of matching human intelligence across all cognitive tasks – is now a matter of “a few years'.

What Google Actually Presented

Beyond the remark about the singularity, I/O 2026 featured a series of concrete announcements. The new generation of the Gemini 3 model was unveiled, with a focus on agentivity – the ability of models to execute complex tasks end-to-end. Gemini Omni, presented by Google in a dedicated video on May 22, is a multimodal model capable of processing text, image, audio, and video, and generating video that can be edited through simple conversation. Google also highlighted “agentic coding,” assistants integrated into Android 17, and a wave of features aimed at transforming search into a conversational interface.

Singularity: Concept, Promise, and Danger

The word “singularity” is not inconsequential. Popularized by futurist Ray Kurzweil in the early 2000s, it designates the theoretical moment when technology irreversibly surpasses human capabilities – a threshold beyond which the world can no longer be understood by our intuition alone. The fact that the leader of one of the world's leading AI labs uses this term publicly constitutes a paradigm shift. However, Hassabis tempers his own statement: “The singularity, at least my interpretation, refers to the era we are entering.” Not an imminent cataclysm, but a period – one we are already living through.

The Race for AGI: Google vs. OpenAI

This statement comes amid a fierce competition between major research labs. OpenAI boasts a stratospheric valuation, its models dominate a portion of consumer uses, and Sam Altman's firm is investing heavily in computing power. Meanwhile, Gemini regularly secures top rankings in open leaderboards like LMArena, with particular strength in multimodality and grounding – the ability to anchor responses in verified facts. The rise of reasoning models, combined with the growth of autonomous agents, is transforming what was once science fiction into operational reality.

“Ten Times the Impact of the Industrial Revolution”

For Hassabis, cited by Business Insider, AGI will have “ten times the impact of the industrial revolution, but unfolding ten times faster – in a decade rather than a century.” A striking statement that reveals the urgency felt by the man behind AlphaGo and AlphaFold regarding what he is creating – and the dizzying responsibility that it entails. The CEO of DeepMind has been publicly emphasizing the need for international governance frameworks for several months – at the risk, he believes, of seeing decisions about the future of humanity “captured by two or three private labs.”

Editorial Opinion

When the man who co-led the creation of AlphaFold – a system that, in a few years, enabled the prediction of the structure of hundreds of millions of proteins – tells you that AGI is “a few years away,” his words must be taken seriously. Not with panic, but with all the gravity the moment deserves. The question is no longer “if” AGI will arrive, but “in what world” it will arrive. A world where governments have considered the governance of these systems, or a world where a handful of private actors have imposed their own rules? OrChair poses this question every day. It does not yet have a satisfactory answer.

Key Takeaways

  • May 19, 2026: Demis Hassabis speaks of the “foothills of the singularity” at Google I/O.
  • AGI predicted by him “in a few years”.
  • Key announcements: Gemini 3, Gemini Omni (multimodal video), AI agents, and Android 17.
  • Hassabis: AGI will have “ten times the impact of the industrial revolution in ten times less time”.
  • Intense race between Google DeepMind and OpenAI on reasoning models and agents.
  • Renewed call for international AI governance.