A Nobel Prize-winning economist is publicly challenging Big Tech’s vision of an AI-powered job revolution, insisting that autonomous agents will not eliminate the need for human labor. In an exclusive interview, Daron Acemoglu of MIT reaffirmed his 2024 estimate that artificial intelligence will deliver only a modest boost to US productivity and will not trigger mass unemployment.
“I think that’s just a losing proposition,” Acemoglu said of the notion that AI agents can serve as one-to-one replacements for entire positions. He argues that these systems are better suited as tools to augment specific tasks rather than handle the full breadth of a job.
Background
Months before winning the 2024 Nobel Prize in economics, Acemoglu published a paper that earned him few fans in Silicon Valley. He estimated that AI would add only a small fraction to US growth and would not make most white-collar work obsolete. His measured prediction has not gained traction: the fear of an AI-driven jobs apocalypse now appears everywhere from Bernie Sanders’s rallies to casual grocery-store conversations.
Some previously skeptical economists have shifted, and a California gubernatorial candidate recently proposed taxing corporate AI use to compensate laid-off workers. Yet Acemoglu points out that employment data still supports his original thesis—studies repeatedly show AI has not altered overall hiring or layoff rates. However, the technology has advanced significantly since his paper, most notably with the rise of AI agents.
AI Agents: The New Frontier
The biggest technical leap since Acemoglu’s paper has been agentic AI—tools that act autonomously to complete user-defined goals. Companies now pitch these agents as a one-to-many replacement for human workers, but Acemoglu calls that a misunderstanding of how jobs work.
“One reason has to do with all the various tasks that go into a job,” Acemoglu said. He has studied this since 2018, using the example of an x-ray technician who juggles 30 tasks: taking patient histories, organizing mammogram archives, and more. “A worker can naturally switch between formats, databases, and working styles,” he explained. “But how many individual tools or protocols would an AI require to do the same?”
The key question is whether AI agents will eventually master the orchestration between tasks that humans perform naturally. AI companies are fiercely competing to show their agents can operate independently for longer periods without errors, sometimes exaggerating results. Acemoglu maintains that many jobs will be spared if agents cannot fluidly switch between tasks.
What This Means
If Acemoglu’s view prevails, the current hype around AI agents could lead to overinvestment and missed opportunities. Companies may pursue full automation where augmentation would be more practical, wasting resources and disappointing shareholders. Workers, meanwhile, may continue to worry about job security despite data showing minimal displacement so far.
Policy makers considering taxes on AI use or mandatory compensation for “AI-driven layoffs” should note that such measures may be premature. Instead, the focus should shift to retraining and upskilling workers to leverage AI as a tool—much like the x-ray technician who uses new software without losing her job. As Acemoglu puts it, “If agents can’t switch between tasks fluidly, many roles remain perfectly safe.”
For now, the economist’s calm, data-driven approach stands as a counterweight to apocalyptic narratives—though he admits he watches developments closely. “I’m not saying nothing will change,” he said. “But the notion that any day now we’ll all be replaced by bots is not supported by evidence.”