IBM CEO Arvind Krishna thinks A.I.’s ability to increase worker productivity is the solution to a vexing problem faced by many developed countries: shrinking workforces.

“In all of these developed countries there is actually a disinflation in the demographics,” he said in an interview with CNBC on August 22. “Population is flat or in the worst case declining. Then if I look at people who are working age, it is declining, so you need to get productivity. Otherwise, quality of life is going to fall. And A.I. is the only answer we got.”

Working-age populations in some of the world’s most important economies such as Japan, South Korea, the U.K., and China are indeed declining, according to the World Bank. In the U.S. the picture was a little rosier, with the numbers of workers remaining flat since 2020 after decades of steady increases—a trend likely exacerbated by the Great Resignation.

Krishna was also responding to the unusually tight U.S. labor market that has made it difficult to hire enough employees. His point was that firms can no longer hire as many new employees as they would like, and therefore must get accustomed to doing more with less, or at least the same resources they currently have.

With A.I., Krishna says, companies “can get the same work done with fewer people—that’s just the nature of productivity.”

He’s convinced humans will work alongside A.I. rather than be replaced by it. “You get digital labor or A.I. bots augmenting or working alongside their fellow humans doing that work,” he said.

In a commentary for Fortune in April, Krishna provided some examples of how he expected that process to play out. Citing the example of human resource departments, he said IBM can now deploy just 50 workers using A.I. to do what previously required 700 employees.

“That’s freed up a very significant number of people to spend more time providing important talent-related services, such as career guidance and support for managers, which requires thought and creativity, rather than doing routine paperwork,” he wrote.

Jobs that require repetitive, manual tasks are the ones that Krishna thinks are first in line to be impacted by A.I. “I actually believe that the first set of roles that will get impacted are, what I call, back office, white-collar work,” said Krishna. And while that raises the specter of mass unemployment and pervasive job cuts, Krishna has said previously that A.I. will create more jobs than it eliminates. Whether his prediction comes to fruition remains to be seen.

An IBM spokesperson said the company had nothing to share beyond beyond Krishna’s CNBC interview and Fortune commentary.

This wasn’t the first time Krishna has spoken about artificial intelligence’s impact on jobs. In May, he said IBM would freeze hiring for 7,800 roles the company thought could be replaced by A.I.

In August 22’s CNBC interview, Krishna said he was “slightly misquoted” in the coverage of those remarks. Rather than eliminating those roles, IBM simply won’t backfill those that are left open through the normal annual attrition of 5% to 6%. Crucially, the policy will be in place only for five years, after which presumably hiring will pick up again, but for a host of new roles created because of A.I.

IBM was among the first companies in artificial intelligence

IBM was among the earliest players in artificial intelligence. Long before generative A.I. became available to the public through services like OpenAI’s ChatGPT chatbot, IBM had created Watson, a machine-learning-powered tool that responded to natural language. The initial promise of Watson—that it could help in health care and professional services—may not have panned out. Customers mostly felt the tool was limited to answering trivia questions, rather than revolutionizing how people work.

But IBM took the experience it gained by developing Watson as a universal application and tweaked its business model to create A.I. tools for specific use cases. To cater to A.I. developers, IBM introduced Watsonxai, a service that provides access to large language models that are the basis for most artificial intelligence technologies. Like many others the company also jumped headfirst into generative A.I. with Watson Assistant, meant to rival ChatGPT and Google’s Bard. Additionally, IBM has an A.I. tool called IBM RXN for Chemistry, which is pitched as being able to predict the outcomes of chemical reactions.

Krishna reiterated on August 22 his belief that A.I. would be universally applied, saying it will make “every enterprise process more productive.”