AI Productivity Gains Leave Employers Debating Worker Benefits

Companies are tracking AI-related time savings and output gains, while Bristol Myers Squibb executive Amanda Poole says a proposed way to share productivity gains with workers remains unresolved.

At an Oct. 12-14 business summit in Santa Barbara, California, leaders from PepsiCo, Indeed, Dropbox and Bristol Myers Squibb discussed ways to measure and use AI gains. Athina Kanioura, PepsiCo's global chief strategy and transformation officer and CEO of Latin America Foods, questioned the pressure to show an immediate financial return for every AI application.

Indeed's AI coding assistants have saved engineers an average of more than four hours a week, while the company measured overall engineering productivity as more than 50% higher with AI tools. Those figures cover engineering work; they do not show how much time employees received away from work.

At Dropbox, 85% of time saved with AI went back into work activities. Forty-nine percent of employees said they directed saved time toward higher-value work. Dropbox Chief People Officer Melanie Rosenwasser said, “time saved does not necessarily equal value created.”

Poole, Bristol Myers Squibb's executive vice president and chief people officer, raised the possibility of dividing a productivity gain between the company and workers through a “well-being quotient.” Her example of a 20% productivity increase was hypothetical, not a company result. “We haven't solved it, so more to come,” Poole said.

The summit's examples used different measures: Indeed tracked engineering hours and output, Dropbox examined how employees used saved time, and Poole discussed how a gain might be shared.

The issue carries implications for workers beyond how they use an hour saved. Across U.S. survey waves from 2023 through early 2026, people who used AI daily or multiple times a week were more than twice as likely in most waves as less frequent users to say it was very likely that technology, automation or AI would eliminate their jobs within five years. That comparison shows an association between frequent AI use and concern, not that AI use caused the fear.