The soaring valuations of AI companies aren't just a bet on better software.
Those are bets about who will control human labor in the future, said Roman Yampolsky, a computer science professor at the University of Louisville, one of the first academics to warn about the risks of AI.
Yampolsky said the market is pricing in a fundamental shift in which machines will provide “free labor” at scale, as artificial intelligence moves from tools to increasingly autonomous agents.
“You go from having tools to having agents with human-like abilities that actually represent free labor,” Yampolsky said in a recent interview with Britain's LBC radio station. “Free Labor — Cognitive Free Labor, Physical Labor.”
This dynamic helps explain why investors are willing to pay high valuations for AI companies, even before many have established clear business models, he said.
“Even if a company is valued at 100 billion today, it's really a small gamble to get that free labor,” he told LBC.
In a subsequent comment to Business Insider, he said, “Once a model is trained and deployed, it's primarily compute, not payroll, that copies its functionality across millions of tasks.”
He added that the market tends to underestimate how sudden change can be.
“Sudden substitution can occur when quality exceeds usability standards,” he said. “Wage values may collapse faster than institutions can adapt.”
Jobs at risk — and why this time is different
Yampolsky told LBC that jobs that run entirely on computers are vulnerable to automation. He cited programming, accounting, tax reporting, and web design as examples.
Automation may initially remove only the most boring tasks, but over time entire roles can disappear, he said.
“Within five years, we will be able to automate all cognitive labor and a lot of physical labor,” he said in an interview, pointing to rapid advances in robotics.
What makes this wave of technology different from others is that AI targets “the general underpinnings of cognitive work itself,” rather than specific tasks, Yampolsky told Business Insider.
Past technologies have created new jobs that require uniquely human skills. The frontier continues to move this time around, he said.
“New categories become automatable as soon as they emerge,” he said.
He expects adoption will be slowed by regulation, liability and organizational inertia, but ultimately driven by competitive pressures and cost reductions.
“Companies will embrace anything that reduces costs and increases speed, as long as it's reliable enough,” he said.
The debate over AI and employment is divided
Yampolsky has previously said that AI could put 99% of workers out of work by 2030, joining a growing chorus of AI experts and technology leaders predicting that the technology could eliminate a huge number of jobs.
Computer scientist Geoffrey Hinton, known as the “godfather of AI”, said “many jobs” could be replaced by AI as early as 2026, and AI pioneer Stuart Russell warned that society could face unemployment rates of up to 80%.
But others disagree.
Nvidia CEO Jensen Huang and former MetaAI chief Yann LeCun say AI will change rather than eliminate jobs, and executives such as JPMorgan's Jamie Dimon and Zoom's Eric Yuan have predicted that the technology could reshape jobs and shorten the workweek rather than eliminate them.
Yampolsky believes there is a greater risk. As society becomes dependent on AI for critical infrastructure, he said, slowing down may no longer be a viable option.
“Addiction creates lock-in,” he told Business Insider.
“As AI becomes critical infrastructure, risk tolerance increases by necessity rather than choice. In that state, the very possibility of control failure increases as the options of pausing, auditing, or rolling back are no longer viable,” he said.
