Most people agree that AI would cause job losses, but some predictions are a lot more dire.
Roman Yampolskiy, AI safety researcher and author of AI: Unexplainable, Unpredictable, Uncontrollable, believes AGI is roughly a year away — and that when it arrives, it will effectively end employment as we know it. Speaking on a recent podcast, Yampolskiy laid out a timeline that is stark even by the standards of a field accustomed to bold claims.
“We’re probably looking at AGI as predicted by prediction markets and the tops of the labs,” he said. “We’d have artificial general intelligence by 2027.”
On what that would mean for the economy, Yampolskiy didn’t hedge: “If you have free labor — physical and cognitive — trillions of dollars of it, it makes no sense to hire humans for most jobs if you can just get a $20 subscription or a free model to do what an employee does.”
He sees the transition unfolding in two waves. “First, anything on a computer will be automated. And next, I think humanoid robots are maybe five years behind. So in five years, all the physical labor can also be automated.”
The conclusion he draws from this is not a softened version of disruption. “We are looking at a world where we have levels of unemployment we’ve never seen before. Not talking about 10% unemployment, which is scary, but 99%. All you have left is jobs where, for whatever reason, you prefer another human to do it for you.”
That 99% figure maps roughly to 2032 — five years after his AGI arrival date — when he expects humanoid robotics to close off the last large category of human work.
The prediction sits at the extreme end of the spectrum, but the broader direction it points toward is increasingly mainstream. Anthropic has said it expects “far more dramatic AI progress” in the next two years, with its CEO Dario Amodei placing AGI arrival in the 2026–2027 window. More strikingly, OpenAI’s VP of Research Aidan Clark recently hinted that AGI may have already arrived — a claim that would have seemed outlandish just a few years ago. Whether or not Yampolskiy’s 99% figure proves accurate, the compressed timeline he describes — cognitive automation now, physical automation by 2032 — is no longer a fringe position. The real question may be less about whether mass displacement happens, and more about how quickly societies can build the structures to absorb it.