Data reveals workers are spending hours fixing confident-but-wrong AI mistakes, leading to a much more dangerous corporate habit.
With estimates of business artificial intelligence adoption often ranging from under 20 percent to more than 90 percent, employers can be forgiven for any confusion about how widely the tech has been deployed—and how effective it has been. Recent research data offers positive details on both topics. But it also sounds a warning that still-developing AI tools are forcing workers to spend considerable time “botsitting” apps to correct shoddy output—or, in some cases, just circulating “botshitting” content throughout the workplace when that vetting work gets old.
Those were top findings in a study by AI tools and platform developer Glean, which surveyed 6,000 digital workers in the U.S., the U.K., and Australia. On the upside, fully 87 percent of respondents said they now use the tech for their work—well to the upper end of most estimate ranges. Better still, an equally impressive 75 percent reported the apps have made them more productive, saving them on average 11 hours each week.
But it’s at that point that things go a bit pear-shaped, to put it politely.
Too much botsitting leads to botshitting
Because rather than using the time economized through work automation for more valuable tasks, survey respondents said they spend an average 6.4 of those hours botsitting. Glean defined that as feeding apps “missing context, checking its outputs, debugging its mistakes, rerunning prompts, and cleaning up the confident-but-wrong answers AI leaves behind.”
In other words, employees use more than half of the time bots save them by scanning for, catching, and cleaning up another term the tech has inspired: workslop.
Think that’s bad? Glean found an even worse consequence of AI use—and a singularly graphic term for it.
It comes into play when the untracked, unbudgeted, and unrewarded grind of botsitting becomes too boring, repetitive, and thankless for workers to bear any more. When that happens, many employees simply pass the unchecked content their AI wonder tools have spat out to the rest of the workplace, or even to customers.
“That’s when botsitting turns into something more dangerous: botshitting—shipping AI-generated work that workers haven’t reviewed, don’t
fully understand, or couldn’t defend if asked,” the report says, before adding another finding employers should worry about. “Today, 69 percent of AI users admit to botshitting at work.”
And you thought Cory Doctorow had a knack for linking unhappy tech outcomes to the final act in the digestive process.
A related problem to those exasperated employees no longer checking AI output is when they also abdicate responsibility for the dreck that drove them to give up in the first place. Nearly 30 percent of respondents said they blame the app for the slop produced, a number that rose to 41 percent among heavy users.
Employer fixes for botsitting and worse AI consequences
So how can business owners respond to these troublesome findings? Glean has two clear recommendations.
The first is for organizations that have been focusing on maximizing staff use of AI to fuel wider adoption to instead return their attention to the employees using the tech. That means prioritizing quality of results, not quantities of times using apps.
That shift involves defining how, when, and why AI is used, then regularly measuring both legitimate increases in output, and the quality of that content. In other words, Glean found successful companies incorporate bots as part of a holistic redesign of workplace roles and activities, then check and tweak that evolving process to improve it.
“They measure what matters, not what’s easiest to count,” Glean said in detailing that effort. “They make governance a living system. They start with the work, not the tech stack. They ground AI in enterprise context, not just enterprise data. They invest in their people.”
The upshot of making those changes is usually a considerable increase in truly productive results, and a lot less employee botsitting and—you know, that sweary thing.
This post originally appeared at inc.com.
“Click here to subscribe to the Inc. newsletter: inc.com/newsletters"