Will artificial intelligence (AI) replace many human jobs soon? Nobel laureate economist Daron Acemoglu says AI may replace only about 5% of human work over the next 10 years. He believes AI should be developed to help workers rather than fully automate their jobs.
- AI could automate about 20% of simple office and cognitive tasks in the US economy.
- Only about one-fourth of these tasks are expected to be automated within 10 years due to slow technology adoption.
- Real-world productivity gains from AI remain modest compared to controlled environments.
- Acemoglu proposes "pro-worker AI" that enhances human skills instead of replacing workers.
- AI can assist teachers and professionals by providing better information and support.
How much work will AI replace in the next decade?
Acemoglu estimates that AI will replace only about 5% of the work humans currently do within the next 10 years. This estimate is based on the capabilities of current AI models and the typical pace at which new technologies are adopted in the economy.
He explains that while AI could automate around 20% of simple office and cognitive tasks, only a fraction of these tasks will actually be automated soon because technology diffusion takes time.
Why are real-world productivity gains from AI modest?
Despite rapid advances in AI models, Acemoglu notes that companies using AI have not seen large productivity improvements. He points out that AI systems perform well in controlled environments but face challenges in real-world settings.
Achieving the reliability and versatility needed to automate entire jobs remains difficult. Many professions such as radiologists, journalists, paralegals, accountants, and office workers continue to be essential despite earlier predictions of rapid AI replacement.
What is "pro-worker AI" and why does Acemoglu support it?
Acemoglu advocates for "pro-worker AI," which means developing AI technologies that complement human workers. Instead of fully automating jobs, AI should improve workers' capabilities, help them acquire new skills, and create new types of tasks.
For example, AI could help teachers identify students' difficulties and reorganize classrooms to provide better support, rather than replacing teachers entirely.
This approach values the differences between human intelligence and artificial intelligence and aims to combine their strengths.
What challenges affect AI's impact on jobs?
The current focus on automation is influenced by business models, concentration in the tech sector, investment incentives, and a strong interest in artificial general intelligence (AGI). Acemoglu suggests that changing perspectives on AI development is key to achieving better outcomes.
He emphasizes that there is no single policy solution but stresses the importance of prioritizing AI that helps workers solve problems and develop new skills.
Frequently Asked Questions
Q: Why does Acemoglu believe AI will replace only 5% of work?
A: He bases this on AI's current capabilities and the slow pace of technology adoption, which limits how quickly AI can automate tasks.
Q: What kinds of tasks can AI automate?
A: AI can automate simple office and cognitive tasks, but fully automating entire jobs is still difficult.
Q: How can AI support workers instead of replacing them?
A: By providing better information, enhancing skills, and creating new tasks, AI can complement human work rather than substitute it.
Q: What is the main obstacle to AI automating more jobs?
A: The main challenges are achieving the reliability and versatility needed for full automation and the slow diffusion of technology in real-world settings.
Source: Article by Nobel laureate Daron Acemoglu published in The Humanist Review, October 2026.
