AI leaders quietly abandon shared definition of AGI

The two most prominent figures in the race to build advanced artificial intelligence are pulling in opposite directions — not just on what that technology should be called, but on how to persuade a skeptical public to accept it.

OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei have staked out contrasting positions on restoring public confidence in AI, with Altman pushing a broad, aspirational vision and Amodei insisting that only concrete, verifiable results will win people over. The split comes as Bloomberg reported on September 7 that the industry's leading figures are quietly abandoning any shared definition of artificial general intelligence, even as they continue to promise it is near.startupfortune+1

A Term in Retreat

Altman now calls AGI "a very sloppy term" and has shifted to talking about superintelligence arriving in "a few thousand days," a phrase he first used in a September 2024 blog post. He told TIME in August that OpenAI would have an internal system he would call AGI by the end of 2026, but OpenAI's own definition — a system that outperforms humans at most economically valuable work — remains self-set and untested by any outside body.the-decoder+2

Amodei, meanwhile, has largely stopped using the word AGI altogether, preferring "powerful AI." In a March 2025 policy filing to the White House, Anthropic told regulators to expect such systems by "late 2026 or early 2027". His working definition is narrower and more concrete: intellectual capabilities matching or exceeding those of Nobel Prize winners across fields from biology to engineering, what he has described as "a country of geniuses in a datacenter".startupfortune+2

Google Alphabet Inc. ↘ 2.09% DeepMind CEO Demis Hassabis has offered yet another timeline, tightening his estimate from roughly a decade to three to five years by early this year. Nvidia's ↘ 0.91% Jensen Huang, for his part, has framed AGI as a spectrum the industry is already climbing — a formulation that lets him claim progress without naming a finish line.startupfortune

The Trust Gap

The definitional disagreement feeds directly into a widening debate about public trust. In an August exchange on X, Amodei acknowledged what he called "fundamentally a crisis of trust," writing that "ordinary people don't trust companies, governments, or the tech industry and always suspect that we are cooking up some new way to screw them over". He conceded that AI companies "haven't yet delivered on our big promises to benefit the world" and endorsed the creation of a FINRA-style independent regulator for frontier AI.fortune+2

Altman has taken a different tack, leaning into optimism and sweeping predictions rather than institutional reforms. Both CEOs have also recently walked back earlier warnings about mass white-collar job displacement as their companies approach potential IPOs.political

Benchmarks Without Borders

As Bloomberg's Austin Carr put it, for a growing number of researchers AGI has become less a testable threshold than something closer to "a spiritual concept" — a horizon that recedes as fast as the industry approaches it. Until a lab commits to a public, third-party benchmark with an actual date attached, every timeline in the debate remains a promise made by the people who profit most from it being believed.bloomberg+1