Why Workers Now Spend More Time Managing AI Than Doing Their Actual Jobs

A new CIO Dive report has put a number on a frustration millions of office workers already feel: of the hours people spend interacting with AI, they now spend slightly more time managing the tools than actually using them to produce work. The finding, drawn from Glean's Work AI Index survey of 6000 digital workers, lands at the heart of the productivity question every job seeker is asking in 2026. At Metaintro, we cover how AI is reshaping real workdays, real job descriptions and real career paths, and this story explains why "AI saved me time" and "AI made me more productive" are turning out to be two very different claims.

What did the Glean survey actually find?

Glean's Work AI Institute surveyed 6000 digital workers and combined those responses with insight from AI leaders who use its platform, publishing the results in its inaugural Work AI Index report. The headline number is striking. Three-quarters of knowledge workers said AI makes them more productive, yet only 13 percent reported that the technology had significantly improved their company's overall performance. That gap between how individuals feel and what organizations can measure is the whole story in one statistic.

On average, automation saved workers about 11 hours a week. That sounds like a clear win until you look at where those hours went. Employees reported spending much of that saved time managing the AI itself. When researchers broke down the hours people spend interacting with AI, workers said they spent slightly more time managing the tools than using them to generate output. Only 27 percent of that time went to learning how to use the tools and building agents, the kind of investment that compounds over a career.

The report frames the problem bluntly. Rebecca Hinds, head of the Work AI Institute at Glean, said too many companies are treating AI adoption like a vanity metric, chasing more seats, more prompts and more usage. But more AI usage does not equal productivity or genuine transformation. That disconnect mirrors a pattern we have tracked closely in our reporting on why 90 percent of companies see zero AI productivity gains, where the enthusiasm at the individual level keeps failing to show up in company results.

Where do the hours actually go?

The most concrete finding in the report is the maintenance number. Workers spent nearly six and a half hours per week on low-visibility tasks that exist only because the AI exists: giving agents context, checking their work, flagging mistakes and cleaning up answers. These are not the tasks anyone lists on a resume. They do not show up in a status update or a performance review. They are invisible labor, the digital equivalent of tidying up after a tool that was supposed to tidy up after you.

It gets sharper. The report found that for every hour an employee spends getting a useful output from AI, they spend another hour making that output usable. That is a one-to-one ratio between extraction and repair. And the failures are not rare. More than a third of AI sessions fail completely, the report found, forcing employees to start over from scratch. When a third of your attempts collapse, the time you "saved" starts to look more like time you simply moved from one column to another.

This is the same dynamic we described in our coverage of the AI productivity trap, where 31 percent of workers report heavier workloads as employers raise output expectations on the assumption that AI is doing the heavy lifting. The tool removes one kind of work and quietly hands you another. The label on the new work is "managing AI," but the experience is closer to supervising an eager intern who never quite learns.

Why does individual productivity not show up in company results?

This is the puzzle at the center of the data, and it is not new. Economists have been circling it for two years. We unpacked it in the AI productivity paradox, where 80 percent of companies see no gains while economists spot a breakthrough, and the Glean numbers add a human-scale explanation. The gains workers feel are real in the moment, a faster first draft, a quicker summary, a head start on a spreadsheet. But those gains leak out through review time, error correction and rework before they ever reach the bottom line.

Successful enterprise AI adoption, the report argues, hinges on the human infrastructure around the tools rather than the tools themselves. Leaders need to ground AI in real company context, train employees on specific use cases, treat unsanctioned "shadow AI" as a signal that approved tools are falling short, and build governance into everyday decisions. In other words, the technology is the easy part. The hard part is the organizational scaffolding, and most companies have not built it yet. That is the same conclusion drawn by the 6000 executives who exposed AI's productivity mirage in a separate survey, and it is why so many AI rollouts stall after the pilot.

There is a structural reason the company-level number stays stuck at 13 percent. When every worker absorbs a few extra hours of cleanup, no single person flags it as a crisis, but the aggregate drag is enormous. Multiply six and a half hours of maintenance across thousands of employees and the "time savings" can vanish entirely at the organizational level, which is exactly what the gap between 75 percent and 13 percent suggests.

Is constant AI oversight quietly hurting your skills?

Here is the part of the story that should worry job seekers most, and it has nothing to do with hours. The report noted that 69 percent of workers say they carefully review AI outputs and verify that the recommendations make sense. That vigilance is the only thing keeping mistakes from slipping through. But the report also warned that monotonous review work easily turns into mistakes of its own, because attention fades when the task is repetitive and the stakes feel low.

There is a deeper cost. When your job becomes proofreading a machine instead of building something from a blank page, the muscles that made you valuable can atrophy. We explored exactly this risk in the AI comfort trap, where automation is quietly eroding the skills that make you hireable and in our look at the critical skills your team is quietly losing to AI. If you spend your days editing AI drafts rather than producing original analysis, your judgment stays sharp but your generative ability can dull, and generative ability is what most employers still pay a premium for.

The split between AI-fluent workers and everyone else is also widening, which raises the stakes on every hour you invest. Our reporting on why 72 percent of workers can't even use their company's AI tools shows that access alone is not the differentiator anymore. The differentiator is whether you use those hours to build durable skills or simply to keep a noisy tool from breaking.

What does the human side of this look like?

The maintenance burden is not just a workflow problem. It changes what it feels like to do a job, and it changes which jobs are pleasant and which are exhausting. That distinction matters enormously, because not all AI at work is created equal. As Lacey Kaelani, CEO of Metaintro, told Nation's Restaurant News, "There's a major difference between AI that assists in scheduling hourly workers, a significant issue, and AI that rates an employee's politeness."

That line captures the divide running through the entire conversation. AI that genuinely removes friction, that smooths scheduling or drafts a routine email, frees you to do higher-value work. AI that monitors, scores and demands constant correction adds a layer of supervision and surveillance that drains the day. The Glean data suggests far too many deployments have landed in the second category, where workers spend their saved hours feeding, checking and apologizing for the machine. The question for any worker evaluating a new role is no longer "does this company use AI" but "what kind of AI work will I actually be doing all day."

What does this mean for your career?

The data points to a clear set of moves for anyone trying to stay valuable as AI reshapes the workday. Managing AI well is becoming a skill in its own right, and the workers who treat it as one will pull ahead of those who treat it as a chore.

• Treat AI oversight as a real competency, not busywork. The 27 percent of time spent learning tools and building agents is the part that compounds. Lean into it. See our guide to the skills that will thrive alongside AI agents in 2026.
• Protect your core craft. Keep producing original work, not just editing AI drafts, so the skills employers pay for stay sharp, as we lay out in the AI comfort trap.
• Get fluent fast. Workers who master prompting and agent-building command real wage premiums, detailed in our piece on the fastest-growing AI skill that isn't coding.
• Know what good AI work looks like before you accept a role, using the agentic AI skills that will define the next wave of hiring.
• Track your invisible labor. If six and a half hours a week is going to cleanup, make that visible to your manager so it counts toward how your performance is judged, as we recommend in our guide to thriving when your employer mandates AI use.

The labor market is rewarding people who can turn AI from a time sink into genuine leverage. With three-quarters of knowledge workers already feeling more productive, the next competitive edge is not using AI more, it is using it well enough that the company actually sees the gain, and so does your next employer.

Related Articles

• AI 'Workslop' Is Flooding Offices, How to Spot It and Stop It
• The AI Productivity Trap: 31% of Workers Report Heavier Workloads as Employers Raise the Bar
• Why 90% of Companies See Zero AI Productivity Gains, and What Job Seekers Should Learn From It
• 6,000 Executives Exposed AI's Productivity Mirage, Here's What It Means for Your Job Security
• The AI Productivity Puzzle: Why 80% of Companies See No Gains While Economists Spot a Breakthrough
• The AI Comfort Trap: How Automation Is Quietly Eroding the Skills That Make You Hireable
• The Hidden Cost of AI at Work: 4 Critical Skills Your Team Is Quietly Losing in 2026
• Why 72% of Workers Can't Use Their Company's AI Tools, and What It Means for Your Career
• Agentic AI and Your Career: 5 Skills That Will Define the Next Wave of Hiring in 2026
• Your New AI Coworker Just Got Hired: 7 Skills You Need to Thrive Alongside AI Agents in 2026
• Your Boss Just Made AI Mandatory, Here's How to Thrive (Not Just Survive)
• Why Corporate Reskilling Programs Are Failing Workers, And What Actually Future-Proofs Your Career in 2026

People Also Asked

Q: How many hours a week do workers spend managing AI?

A: According to Glean's Work AI Index survey of 6000 digital workers, employees spent nearly six and a half hours per week on AI maintenance tasks like giving agents context, checking their work, flagging mistakes and cleaning up answers. Of the total time spent interacting with AI, workers reported spending slightly more time managing the tools than using them to produce actual work.

Q: Does AI actually make companies more productive?

A: The picture is mixed. Three-quarters of knowledge workers say AI makes them personally more productive, but only 13 percent report that it significantly improved their company's overall performance. The gap exists because individual time savings leak out through review, error correction and rework, and because most companies have not built the training and governance needed to convert AI usage into measurable results.

Q: Is relying on AI bad for my career skills?

A: It can be if you let it. When your job shifts from creating original work to proofreading machine output, your generative skills can dull even as your editing stays sharp. Most employers still pay a premium for original analysis and judgment, so the safest move is to keep producing your own work, invest time in learning to build and direct AI agents, and treat AI oversight as a skill you are deliberately developing rather than a chore you tolerate.

Future-proof your career by getting ahead of the AI shift instead of cleaning up after it. Metaintro tracks how AI is reshaping every job and matches you with roles where your skills actually compound, so the hours you put in build a career instead of just maintaining a tool. Create your free Metaintro profile and start building the AI-fluent edge employers are paying for in 2026.