As HR has recently taken the helm of AI transformation at many organizations, driving adoption among employees has increasingly become a top priority. But as many workforces have now deployed AI across their practices, the challenge is shifting—it’s no longer just about getting workers to use AI, but doing so in an efficient and effective way that captures its potential.
A new report from Glean highlights the predicament facing HR and focuses on a phenomenon Glean calls “botsitting.” The good news is that employees are incorporating AI to drive efficiencies in their work. About 87% of the 6,000 digital workers surveyed are using AI at work, and about three-quarters say it can make them more productive, saving them about 11 hours a week. Yet, just 13% say the use of AI has improved their organization’s performance, as the productivity gains are often offset by losses, as workers supervise, and sometimes correct, the work of their AI tools.
Researchers define botsitting as: “the work required to make AI usable, including feeding it missing context, checking its outputs, debugging its mistakes, rerunning prompts and cleaning up the confident-but-wrong answers AI leaves behind.”
Eva Spatz, vice president, head of people experience at Staffbase, tells HR Executive that botsitting is “fundamentally an HR issue.” It stems from a lack of trust and psychological safety, not a “software glitch.”
“If employees lack a clear purpose, or worry that an AI mistake could cost them their jobs, they get trapped acting as rigid, hyper-vigilant operators rather than impact architects,” she says.
And this is increasingly eating up employees’ time: The Glean report found that workers are wasting almost an entire workday—6.4 hours—on botsitting each week.
Looking at the total amount of time employees interact with AI in a given week, about 37% is spent botsitting, compared to 36% using the tool to produce work. In other words, for every hour a worker spends on getting usable output from AI, they spend just as much addressing its problems, according to the report. They also spend more than one-quarter learning and building AI agents.
“Most botsitting is grunt work, such as reloading context into different tools, catching hallucinations and verifying outputs that sound confident, or, worse, flatter workers with the answers they wanted to hear instead of what’s true,” researchers write.
A “small share” of this can be productive, but even so, it has a talent cost. The work is “often invisible, unbudgeted and unsupported. Workers who absorb it without recognition or reward grow exhausted. Then they grow resentful. Then they start polishing their résumés.”
Glean found that the more time employees spend iterating with AI, the more likely they are to feel worn out by that work. That exhaustion has many looking to the door: Those who report spending the most time botsitting are significantly more likely to be on the job hunt.
Building the ‘human infrastructure to stamp out botsitting
It’s going to be up to HR to counter those risks by resisting the pressure to pursue AI transformation without real intentionality around the human impact.
“Organizations must build the human infrastructure—not just the technology infrastructure—that makes AI worth using, or they’ll keep paying the bill,” researchers say.
People-centric leadership that equates psychological safety with operational infrastructure is critical, Spatz says. This involves redefining roles in a way that emphasizes “human judgment over raw output,” while honing in on the cognitive anxiety behind botsitting.
When organizations, led by HR, create a “clear human-in-the-loop blueprint” that bakes human oversight into work processes and decision-making, employees can view AI integration as a “healthy partnership,” one in which they act as “critical editors rather than stressed-out machine caretakers.”
That message should be reinforced with manager training to help spot red flags, such as the “audit tax” when employees are working longer hours, which could mean they’re double-checking AI-assisted work. They should also be trained to keep an eye out for employees withdrawing socially from strategic discussions or over-relying on AI for decision-making
Assessing AI rollouts with botsitting in mind—measuring behavioral, qualitative and emotional metrics—Spatz says, can provide realistic insights on the true value of the project.
“If an employee saves five hours a week using AI but fills it with five more hours of automated administrative noise, the rollout has failed,” she says. “True productivity eliminates friction and frees cognitive bandwidth for work that matters most.”
Credit: Source link
