Goldman Sachs Estimates AI Could Affect 300 Million Jobs Worldwide

Economists at the investment bank say programs like ChatGPT could take over a quarter of the work now done in the United States and Europe, hitting office and administrative jobs hardest, while also raising overall economic output.

Economists at Goldman Sachs published a report this week estimating that the new generation of AI programs could expose the equivalent of 300 million full-time jobs worldwide to automation. In the United States and Europe, they estimate, about two-thirds of jobs involve some tasks the programs could do, and for roughly a quarter of all work, the programs could do most of it.

The report is careful with the word "expose." It does not mean 300 million people will lose their jobs. Most jobs, the economists write, will be partly automated rather than eliminated, with people spending their time on the parts a program can't do. And the history of technology is that new work appears — most of the jobs that exist today did not exist in 1940. The report's own estimate is that AI could raise global economic output by 7 percent over a decade.

But the pattern of who is affected is new. Earlier waves of automation hit factory floors and farms. This one, by the report's reckoning, lands hardest on office and administrative work — the jobs that involve reading, writing, and organizing information — and lightest on construction, maintenance, and cleaning. Lawyers are more exposed than plumbers.

Estimates like this are educated guesses about a technology that is four months old in its current form, and they have been wrong before. What makes this one notable is who is making it. A year ago the question of whether AI would take jobs was a debate among futurists. This week it's a research note from a bank.