'Botsitting' Burns Six Hours a Week as AI Adoption Backfires
The News
A report from Glean, covered by HR Executive on 9 July, surveyed 6,000 digital workers and found that 87% are using AI at work, saving approximately 11 hours per week. Yet only 13% said AI has improved their organisation's performance. The gap is explained by what Glean calls "botsitting" — the work of making AI usable, including feeding it context, checking outputs, debugging mistakes, and cleaning up confident-but-wrong answers. Workers spend 6.4 hours per week on this, representing 37% of their total AI interaction time, roughly equal to the 36% spent producing actual work. Glean found that employees who spend the most time botsitting are significantly more likely to be job hunting. Eva Spatz of Staffbase told HR Executive that botsitting is "fundamentally an HR issue" rooted in a lack of trust and psychological safety.
Why It Matters
This is the most useful correction to AI adoption hype published this week. TA leaders have been told that AI tools will save recruiters hours of sourcing and screening time. The Glean data suggests a more uncomfortable reality: those hours may be saved, but a comparable number are being consumed by the invisible work of babysitting the output. For recruiting teams specifically, this matters because the "time saved" metric that justified the AI tool purchase may be illusory if nobody is measuring what the time gets filled with. The finding that heavy botsitters are more likely to be job hunting adds retention risk to the adoption equation — your most AI-fluent employees may be the most burned out by the tools you gave them.
Taking Action
All action is situated in your corporate context. These are starting points, not prescriptions — choose what fits.
Connect
Who to talk with in your enterprise
Your HRIS or people analytics lead: Ask whether your organisation is measuring AI adoption outcomes beyond usage rates. If you're tracking logins and prompts but not time-on-correction or output quality, you have an adoption metric, not an outcome metric. Your head of learning and development: Discuss whether your AI training covers prompt iteration and output verification skills, or just tool activation. The Glean data suggests the gap between "using AI" and "using AI well" is where the hours leak.
Inspect
What to take a closer look at
Survey your TA team informally: for every hour they spend using an AI tool, how long do they spend verifying or correcting its output? If the ratio approaches 1:1 as Glean found, your ROI calculation for that tool needs revisiting. Also check whether your team's overtime or after-hours activity has increased since AI adoption — that's the "audit tax" Glean describes.
Reflect
What to think about more deeply
If an employee saves five hours a week using AI and spends five hours fixing what it got wrong, has the organisation actually improved — or just redistributed the same work across a more frustrating medium?