There’s a new kind of task taking over employee desks: AI babysitting. Lovingly referred to as botsitting, workers are increasingly spending their time monitoring and redoing work performed by their AI tools to ensure that errors and creative hallucinations aren’t slipping into workplace results. This added task of AI oversight at work doesn’t necessarily look like hard work from the outside, but there is a distinct possibility of it detracting from productivity rather than adding to it. This AI productivity paradox is slowly gaining notoriety at work for complicating operations in discrete and thus-far unquantified ways, making it a matter that requires greater reflection.
With employees now spending over 6 hours a week monitoring AI outputs, botsitting appears to be the latest addition to their workplace routines. (Image: Pexels)
Botsitting Diaries: How AI Babysitting Duties Are Taking up Time in the Workplace
Glean’s Work AI Institute recently conducted a survey of 6,000 full-time digital workers across the USA, Great Britain and Australia, and they identified two forms of AI-centric trends making an appearance. One, botsitting, involves the rigorous review of AI tools and services, monitoring their outputs and fixing errors. Employees spend nearly 6.4 hours per week on such babysitting of AI. Much like the constant supervision a new hire requires in their tasks, AI tools demand that someone keep close watch on their performance. Unlike a novice, however, AI tools often require repeated instructions to stick to the task, with limited long-term memory to help them learn from their mistakes.
The survey also identified another problematic consequence of AI usage, which they referred to as “botshitting,” where workers shipped AI-generated work without review, comprehension, or ownership of the work. Reportedly, 69% of workers admit to using AI results without checking or understanding their results fully. The phrase “AI workslop” is now freely used online to discuss poor-quality AI-generated data, and it’s no one surprise that an equally bold word is now available for those who release such content without checking it for flaws first.
The AI Productivity Paradox Blurs the Lines Between Time Spent and Time Gained
With 87% of digital workers using AI and most saving approximately 11 hours of their week, the technology is useful to the workplace in some way, but its downsides don’t receive as much attention or review day-to-day. Very often, AI tools are trained on a large variety of data. However, for niche tasks and company-specific directives, there is still considerable need for employees to find and input the data into the tool. Workers also have to often switch between AI tools for different purposes, leaving them with an incongruent array of platforms they have to order and make sense of.
While this hidden cost of AI continues to rise, we also cannot advocate for less time spent on botsitting, as this supervision and monitoring are critical with AI use. Unchecked, unverified use of AI data can lead to more serious problems in time, forcing entire teams to sit down to fix successive mistakes that arise from the use of that generated data. Babysitting AI is a non-negotiable part of utilizing the technology and should remain standard practice to ensure it operates as intended. What organizations can change is how they build systems and workloads around the tech.
AI Oversight At Work Is Mandatory, but We Can Improve How We Approach the Technology
The illusion of productivity surrounding AI might lead employers to increase employee workloads, as they can now complete tasks faster. However, this could result in employees feeling more burned out than ever, overburdened by the monitoring of AI tools. This will inevitably lead to some employees taking shortcuts and submitting work that has not been verified. With AI automating the part of their roles they are trained for and leaving merely managerial work to them, it is also evident that technology is leaving us with the tasks that aren’t appealing for most to perform.
Studies like the one conducted by the Work AI Institute are helping with the identification of problems surrounding the use of AI, and by paying attention to the data and updating systems, both employers and employees can navigate the technology more easily. Botsitting might be standard practice today, but that does not mean that work routines can’t be improved to acknowledge this responsibility and create allowances and strategies for this phase of work.
As organizations continue to invest in AI, they must equally invest in what the researchers call the “human infrastructure of AI.” The technology is set to transform the world as we know it, but it isn’t beneficial to look forward to the advancements while ignoring their limitations.
Share your experience with “botsitting” in the comments with us. Subscribe to The HR Digest for more insights on workplace trends, layoffs, and what to expect with the advent of AI.
