AI Troubleshooting Productivity Costs Workers a Month a Year

Your AI assistant eats 87 minutes of your workday

If you have spent any time this year fixing an AI tool's mistakes, you are not an outlier. You are the average worker. According to a new study from BambooHR, employees now spend about 87 minutes a day using AI at work, which works out to 22,526 minutes a year, or roughly 47 eight-hour workdays. That is more than two months of your job spent inside chatbots and copilots.

Here is the part that stings: AI troubleshooting productivity now eats 42% of all that AI time, fixing errors and rewriting prompts, while only 35% goes to work that actually moves anything forward. Do the math and you get almost 20 full workdays a year lost to babysitting the assistant instead of doing the job. Workers log on to AI for time savings, 58% say that is their main motivation, and then spend most of the session cleaning up after it.

What makes this weirder is how upbeat everyone remains. Despite the grind of constant corrections, 65% of workers say they feel confident and enthusiastic about using AI at work. People genuinely like the tools. They just quietly lose weeks of the year to them.

Where all those hours actually go

BambooHR released the study, called Redesigning Work: AI's Performance Review, on September 1, and based it on more than 1,600 full-time, salaried workers in the US, plus a subgroup of 520 HR professionals. The numbers paint a picture of a workforce that adopted AI fast and never really stopped to learn how to run it.

Troubleshooting is not the only drain. Knowledge transfer, the way people learn how a company actually works, is shifting away from colleagues. Thirty-five percent of workers say knowledge transfer at their organization now runs primarily through AI tools rather than people. That might sound efficient until you remember that these are the same tools spitting out wrong answers 42% of the time.

The line between work AI and personal AI has basically dissolved. Sixty-two percent of workers admit to using company-provided AI tools for personal purposes, including 17% running a side business or freelance gig and 27% searching for a new job. Going the other direction, 59% have used personal AI accounts for work tasks, and among that group, 71% entered client data, proprietary strategy, or other sensitive company information into tools their employer cannot see. The AI productivity experiment is running partly on shadow IT.

Nicole Csizar, senior director of HR services at BambooHR, put it plainly in the release: workers love being able to ask a question and get an instant answer, but there is a difference between knowing something and developing the judgment to know what to do with it. She said organizations need to be intentional about making sure efficiency doesn't come at the expense of mentorship and development, because those human conversations are where growth happens.

Companies are spending more and managing less

AI tool budgets have climbed at 63% of organizations, according to the report. But adoption is outpacing most companies' ability to manage the costs, in both money and wasted hours. HR departments have updated just 43% of employees' job descriptions on average to reflect AI expectations, even though 75% of workers say AI has already changed how they actually do their jobs. Seventy-seven percent of HR professionals say they plan to update those descriptions within the next 12 months.

Some HR teams are already peeking over shoulders. Eighty-nine percent of the HR professionals surveyed said they actively reviewed employees' AI prompt histories, and 80% admitted to finding biases or problems in the AI systems supporting their work, mostly model bias and demographic bias. Your chatbot sessions might be performance-review material before anyone tells you.

The BambooHR findings line up with other recent data. A Deloitte survey of 25,000 UK workers, reported by LinkedIn News on September 18, found that 63% had used generative AI for work, yet about half said it saved them no time at all. One in three were using AI without their employer's knowledge, and only 28% of organizations had clear AI policies. Two separate studies on two continents are saying the same thing: access to AI is nearly universal, and the time savings are not.

So what actually fixes this

Most people were handed AI with no training and told to figure it out, the same way offices adopted spreadsheets in the 90s and then spent a decade fixing spreadsheet mistakes.

There are practical moves that follow from the data. Companies could spend less on new AI licenses and more on teaching people how to use the ones they have, especially since workers are paying for AI tools themselves while productivity lags. Clear policies on what can go into a chatbot would stop the shadow-data problem, since 71% of workers using personal AI for work are feeding it client information. And managers could treat prompt iteration as a real skill to train, because the AI troubleshooting productivity drain looks like a training gap more than a tool failure.

Until that happens, the 87-minute daily habit is not going anywhere. Workers spend the equivalent of 47 workdays a year on AI because the tools are genuinely useful and genuinely unreliable at the same time. The month you lose to troubleshooting is the price of using an assistant that still needs a babysitter. The companies that train their people first will be the ones that actually get the time back.

Sources: BambooHR, "Redesigning Work: AI's Performance Review" (GlobeNewswire, Sept 1, 2026); WorldatWork, Workspan Daily News Bytes for Sept. 4, 2026.