Google DeepMind CEO Demis Hassabis said at Google's developer conference last week that humanity is standing in the "foothills of the singularity." Society, he said, is running short on time to prepare for AGI.

Why it matters: AI leaders have warned for years that artificial general intelligence could arrive. What's changing is the shrinking timeline and the increasingly urgent tone from people building it.

Driving the news: Speaking with Axios after his Google I/O appearance, Hassabis said his view that AGI could arrive within four years — or sooner — reflects growing confidence that leading labs are on the right technical path.

• "We can see agents really happening now and imagine what they will be in another year, and how useful they'll be," he said.

The big picture: Hassabis still sees 2030 as his central estimate for AGI, but now says 2029 is possible.

• He said the next wave of AI agents should be seen as a societal stress test for far more powerful systems still to come.
• "You can imagine the agentic era in this next year is a little bit like a practice run," he said.

Case in point: The recent controversy around Anthropic's Mythos model, he said, showed that governments and companies are still struggling to keep pace with frontier AI systems.

• "It was probably a good warning shot across the bow," Hassabis said.

Between the lines: Hassabis said he chose his words to provoke more urgency among governments, economists and the broader public to prepare for increasingly powerful AI.

• "This is partly why I use some of the terms I used, yeah, which were a little bit provocative," he said.

The federal government's tentative steps toward reprioritizing safety are a step in the right direction, he said, referring to a potential AI executive order that would mandate testing before new models are released.

• "I think [safety] needs to be accelerated," he said. "This is a good moment to kind of strike while the iron is hot."
• Hassabis said he has some ideas and is talking with leaders at other top AI labs, though he declined to offer specifics.

Yes, but: Hassabis worries the conversation around the society-reshaping impact of AI remains largely confined to tech circles.

• "You've got to take this seriously," he said. "My economist friends, I feel, are still not taking this seriously enough."
• "That needs to change," he told Axios.

Zoom in: One looming milestone is recursive self-improvement — systems capable of materially accelerating their own development.

• "All the leading labs are quite focused on that," Hassabis said. "There'll be clear gains in terms of speed of your research. But there are also risks with that type of system."
• We're not yet at the point where the systems are getting better on their own, but the pace of development is clearly accelerating.
• "I think what we're seeing is soft self-improvement, in the sense of these coding agents are making engineers much more productive," he said.

What we're watching: Whether society makes good use of the few years between now and AGI as time to prepare or just time for a few more cycles of hype and backlash.

A new AI startup tells Axios that proofs created by its algorithms have now been published in several peer-reviewed academic journals.

Why it matters: AI proponents have for years been saying that the technology would lead to scientific breakthroughs, and now that appears to be happening.

Driving the news: Axiom Math (no relation to Axios) says proofs produced by its technology — with human-authored articles — have now been accepted by five leading journals.

• They also have more papers under review at journals, eight papers on arXiv and six more in the pipeline.

How it works: Axiom's tool — called AxiomProver — produces full, machine-checkable proofs in a formal language known as Lean.

• Researchers feed it a natural-language problem statement and the model translates the problem into Lean and develops a proof, with a separate checker verifying every step.
• Once the problem is solved, human mathematicians pair the formal proofs with an academic explanation.
• In some cases the system has been given an open research problem and, over roughly 24 hours, autonomously produced a complete, machine-verified proof, according to Ken Ono, Axiom's founding mathematician.

Yes, but: Large language models, initially mocked for making errors with basic arithmetic, have been making rapid progress in mathematical tasks.

• Last week, OpenAI announced that a forthcoming, general-purpose model solved the planar unit distance problem posed by Paul Erdős in 1946.

What they're saying: Ono characterized OpenAI's discovery as an "exciting development," characterizing Axiom's work as "complementary, but different."

• "We are focused on producing journal mathematics in which human-readable proofs are paired with machine-checkable Lean formalizations and proofs produced by AxiomProver," Ono said.
• "That combination — a conventional mathematical paper together with formal proof certificates — marks an important new step for the field," he added. "To our knowledge, Axiom is the first to bring this model into the journal literature in this explicit way."

The big picture: Axiom said in March that the company raised a $200 million funding round at a $1.6 billion valuation. It's now among a handful of startups trying to produce AI systems focused on being able to solve math problems in ways that are provably true.

• Another is Harmonic, backed by Robinhood founder Vlad Tenev. In January, Nvidia joined a $120 million Series C round that valued it at $1.45 billion.

Google's overhaul of the search bar this week washes away one of the last vestiges of the internet's halcyon era — when search tools felt empowering, social media and swiping were novel, and popular disillusionment had yet to set in.

Why it matters: The AI era and the TikTok-ification of social media have produced a digital world that's responsive to the market demands of today and unrecognizable from a decade ago.

In heralding the "biggest upgrade to our Search box in over 25 years," Google acknowledged that search was a model of the past and that its core product had to be more aligned with the AI age.

• The blue links that colored the Google user experience for decades have been demoted to a secondary offering as zero-click answers get top billing.
• Publishers, SEO firms, affiliate marketers and review sites built around Google traffic are now fighting for visibility in a search experience that delivers fewer clicks.

Keep reading.

• The bulk of SpaceX's $28.5 trillion IPO filing comes from the company's AI business, estimated at a potential $26.5 trillion. (Axios)
• Extensive flooding in Mason County stemmed in part from data center construction. (WSAZ)

Mady here again after finishing a binge watch for the latest season of "Couples Therapy," a show that features real couples receiving real counseling.

The season features a ChatGPT mention after one of the people in therapy, Jason, says he uses the tool for support ranging from fitness advice to other more emotional topics.