Workers Spend Hours Babysitting AI Instead on Valuable Work

Artificial intelligence is helping workers finish old tasks faster, but it is also creating a new layer of work inside companies, according to a new report cited by the Los Angeles Times.

Paul Leonardi, Duca Family Professor of Technology Management at UC Santa Barbara, said many people do not realise how much time they spend managing AI tools before getting the productivity gains they claim.

The report said most employees now spend more than six hours a week “babysitting” their work chatbots. Of the total time workers spend interacting with AI each week, 37% goes to bot sitting, while 36% is spent actually using the tool to produce work.

Leonardi is one of the co-authors of a new study published by the Work AI Institute. The institute’s contributors include academics from Stanford University and UC Berkeley, and it is sponsored by AI company Glean.

Productivity Gains

The research surveyed 6,000 digital workers across the United States, the United Kingdom, and Australia between December and January.

The report found that workers are seeing major personal productivity gains from AI. Around 75% of individuals said AI had improved their productivity.

However, companies are not seeing the same level of business benefit. Only 13% of organisations said they had recorded major business gains from AI adoption.

Hidden AI Work

Leonardi said the productivity boost can lead to wasted time because workers spend many hours correcting AI output and gathering the right files, documents, and internal knowledge needed to make the tools produce useful results.

It described this as a thick and mostly invisible layer of human labour that keeps AI systems working.

For every hour a worker spends getting useful output from AI, they spend roughly another hour making that output usable.

AI Failures

A major reason workers spend so much time managing AI tools is that the systems often fail to deliver usable results.

Workers reported that more than one-third of AI sessions fail, requiring a full restart or major rework.

The report also found that 41% of workers sometimes submit AI-generated work that they would not be able to explain if asked.

Workers as AI Managers

Leonardi said many companies are now expecting individual workers to act like managers of AI tools. He said employees are being asked to manage AI tools and AI agents while also being expected to produce more work.

However, companies are not always accounting for the time and effort required to manage those systems. The report suggests that this issue is unlikely to disappear soon as AI becomes more common in workplaces.