You adopted AI tools to get hours back. New research suggests a good chunk of those hours is quietly disappearing anyway, just into a different kind of work. Researchers have a name for it now: botsitting. And it’s the unglamorous labor of feeding an AI tool context, checking its output and fixing what it got wrong.
If you’ve ever caught yourself re-reading an AI-generated draft three times before you trust it, you already know this cost personally. The good news is that once you can see it, you can manage it.
The Time Savings Were Real, But So Is the Catch
AI is genuinely saving people time. A global survey of nearly 12,000 employees found that 42% of regular AI users save a full workday or more per week using it, a substantial gain by any measure. The catch is what happens next.
The same research found that 66% of those workers get little or no guidance on what to actually do with the time they free up, and 47% said they now spend more time managing and directing AI tools than they used to spend doing the underlying tasks themselves. The time savings exist on paper. Whether they translate into anything useful depends entirely on what happens after the AI hands back its answer.
What ‘Botsitting’ Actually Looks Like
A separate study tracking 6,000 full-time workers put a number on that in-between labor. White-collar employees spend an average of 6.4 hours a week botsitting, the unpaid, unrecognized work of feeding an AI tool missing context, checking its outputs, debugging its mistakes and re-running prompts until something usable comes out.
That labor doesn’t show up on any dashboard, which is exactly the problem. Only 13% of workers in the same study said their organization was performing significantly better because of AI, despite 87% using it regularly. The tool is working. The system around the tool often isn’t.
There’s a real cost to ignoring this too. Workers who spend an outsized share of their AI time botsitting are 73% more likely to be actively job hunting, a signal that unmanaged AI labor doesn’t just waste hours; it quietly erodes morale.
Not Just a Time Problem, It’s an Emotional One Too
Botsitting isn’t the only hidden tax on AI adoption. A global study on workplace AI use found that 42% of employees who use AI regularly say doing so feels like cheating, and that discomfort is even more pronounced among younger workers, with half of Gen Z employees reporting guilt about their AI use.
That guilt has a practical consequence. More than a third of workers in some surveys admit to hiding their AI use from their employer altogether. If people are quietly ashamed of a tool they’re supposed to be using confidently, they’re far less likely to develop the judgment needed to use it well, which means more botsitting, not less, as they second-guess outputs they don’t feel entitled to trust.
Why This Is a Solvable Problem, Not a Verdict on AI
It’s tempting to read these numbers as proof that AI overpromises, but that’s not quite what the data shows. The 42% of workers saving a full workday didn’t imagine that gain, and the tools genuinely are capable of the output people expect from them. What’s missing isn’t the technology, it’s the layer of habit and structure around how people use it.
That distinction matters because it means the fix is within your control, not something you’re waiting on a better model to solve. Every early technology shift has gone through a version of this same lag: The tool arrives, the surrounding workflow takes longer to catch up and the people who close that gap first get the real advantage. Botsitting is a symptom of that lag, not a permanent tax.
Treat the current moment as the transition period it actually is. The organizations and individuals figuring out how to cut botsitting time down now, through better templates, clearer prompts and honest conversations about where the hours are going, will be the ones who actually capture the productivity gains everyone else is still chasing on paper.
How to Cut Your Own Botsitting Time
Start by treating AI output the way you’d treat a first draft from a junior collaborator, not a finished product and not something to distrust entirely either. Spend your review time on the parts of the task that actually require judgment, not re-verifying everything from scratch out of habit.
Build reusable context instead of re-explaining yourself every time. If you find yourself typing the same background information into a prompt repeatedly, save it as a template or a standing document the tool can reference, which cuts a meaningful chunk of the setup work that counts as botsitting.
Set a personal rule for when to stop checking. Decide in advance how much verification a given task actually warrants, a quick scan for a low-stakes email, a careful read for anything client-facing, so you’re not defaulting to maximum suspicion on every output regardless of what’s at stake.
If You Manage a Team, Close the Guidance Gap
Give people explicit permission to use the time AI saves them, and be specific about what that permission looks like. Vague encouragement to “use AI more” doesn’t help. A clear direction to redirect saved hours toward strategic work, client relationships or deep work does.
Normalize AI use openly rather than letting it stay something employees quietly hide. If more than a third of your team feels the need to conceal their AI use from you, you have a trust problem that will cost you more than any productivity gain AI could offer in the first place.
Track botsitting the same way you’d track any other hidden cost. A short, honest team conversation about where AI review time is actually going will surface fixable problems, redundant tools, unclear prompts, missing templates, faster than any dashboard will.
Featured image from PeopleImages/Shutterstock
