Demis Hassabis, the CEO of Google DeepMind and one of the most credentialed figures in artificial intelligence research, used his closing remarks at Google I/O's keynote presentation to describe the current moment as potentially the "foothills of the singularity," according to The Verge. He framed the occasion as a "profound moment for humanity," invoking the prospect of artificial general intelligence and its potential benefits for the world.
The word "foothills" is doing considerable work in that sentence, and it deserves unpacking. The singularity is a concept with a long and contested history, originating in mathematics and physics before being popularized in AI discourse — most prominently by figures like Vernor Vinge and later Ray Kurzweil — to describe a hypothetical point at which machine intelligence surpasses human cognitive ability so completely and so rapidly that the future becomes, in a technical sense, unpredictable. For decades it occupied a space somewhere between speculative fiction and fringe futurism, the kind of idea that serious researchers were careful to keep at arm's length in public. That Hassabis, a man who holds a PhD in cognitive neuroscience, co-founded DeepMind, and accepted a Nobel Prize in Chemistry for AI-driven protein structure prediction, would reach for this language at a major product keynote is not a small thing.
It matters who is saying this. Hassabis is not a tech evangelist or a venture capitalist selling a vision. He is, by any reasonable measure, among the handful of people on earth most qualified to assess where AI research actually stands. He has also historically been measured in his public statements, expressing optimism about AI's potential while simultaneously calling for caution and international coordination on safety. His willingness to place the current moment in proximity to singularity language — even with the hedging softness of "foothills" and "might be" — represents a notable shift in register, or at minimum a deliberate choice to signal ambition at a moment of fierce competitive pressure.
That competitive context is impossible to separate from the setting. Google I/O is a product conference, and the AI landscape it now inhabits looks nothing like it did even two years ago. OpenAI, Anthropic, Meta, and a constellation of well-funded startups are all competing for developer attention, enterprise contracts, and cultural authority over what AI means and where it is headed. Google, despite having arguably invented much of the foundational research that made the current generation of large language models possible, spent a period visibly scrambling to assert leadership in the public conversation. Hassabis closing the keynote with language about AGI and the singularity is, on one level, a research chief making a scientific assessment. On another level, it is a signal to developers, investors, partners, and rivals that Google DeepMind believes it is playing for the largest possible stakes, and that it intends to be taken seriously as the entity best positioned to navigate them.
The likely consequences of this kind of language fall into several categories. For the research community, it will sharpen an already active debate about how responsible it is to invoke singularity framing in mainstream settings. Critics will argue that such language, even qualified, inflates public expectations, distorts policy conversations, and lends credibility to more extreme predictions that have historically hampered serious safety work. Supporters will counter that understating the pace of progress carries its own risks, leaving regulators and institutions flat-footed. Both positions have merit, and this suggests the debate will not resolve cleanly.
For regulators and governments, remarks like these from a figure of Hassabis's stature add urgency to conversations that are already moving faster than most legislative bodies are comfortable with. The European Union's AI Act is already in implementation. In the United States, federal AI policy remains fragmented. Invocations of AGI proximity from credible researchers tend to accelerate calls for oversight, though the form that oversight takes varies enormously depending on who is listening.
For the broader public, the more immediate consequence is likely a continuation of a pattern: AI announcements are arriving faster than the cultural and institutional frameworks for understanding them. "Foothills of the singularity" is a phrase engineered to produce a sense of altitude, of approaching something vast, without committing to a timetable. That ambiguity is intentional and, the likely reading is, strategically useful.
What to watch for next is whether other leading researchers and executives begin adopting similar language, or whether Hassabis's remarks provoke a public pushback from within the scientific community. Also worth watching is how Google's actual product releases over the coming months match the scale of the rhetoric — keynote language sets expectations, and those expectations eventually collide with reality. And perhaps most importantly, observers should watch whether safety and alignment frameworks at DeepMind and its peers are expanding at a pace that even remotely matches the ambition being announced. At the foothills of anything large, the question is always whether anyone has checked the weather higher up.