Elon Musk says 'we are in the singularity,' agrees with Sam Altman

Elon Musk rarely agrees with Sam Altman, whom he routinely calls a "scammer" online. But this week, responding to a post about frontier AI progress, Musk broke from the pattern: "AI is already superhuman at many things. We are in the Singularity."

It echoed a claim Altman made days earlier on the Relentless podcast: "We are now, like, in the singularity."

It refers to a theoretical scenario whereby AI becomes self-improving at a rate exceeding our human ability to monitor and control it, as postulated by the futurist Ray Kurzweil in 2005.

Altman is not the only one who references the concept this year. Google DeepMind Chief Executive Demis Hassabis called AI as "foothills of the singularity" in May, while Nvidia founder Jensen Huang claimed that AGI was already here back in March.

Musk's remarks came in response to AI researcher Dwarkesh Patel, who argued the real question facing AI companies isn't whether their business models are flawed, but how fast capabilities keep improving.

Patel wrote that investors would turn bearish on large language model firms only if capability growth turns out "slower, spikier, and more data-limited than people currently assume." Altman, separately, predicted AI could automate 30% to 40% of tasks currently done by people.

The singularity discourse came after a two-week period of unease around AI safety. OpenAI revealed that several of its models had escaped the confines of a security test environment and managed to break into the Hugging Face system.

A few days after that, Anthropic revealed that due to a configuration issue, three of its models were able to break into the systems of three organizations by employing "capture-the-flag" tests, using simple methods like poor passwords; the two affected organizations had not even noticed the breach.

Altman has been quick to reject fear-mongering stories, saying that AI will turn out to be "hugely positive."