The ‘Botsitting’ Burden: Why UK Firms are Failing to See AI Productivity Gains
Table of Contents
The Hidden Overhead of the AI Office
For years, the corporate promise of generative AI has been a frictionless leap in productivity—a world where hours of grunt work vanish into a few seconds of processing. However, new data suggests that for the UK workforce, this “dividend” is being spent on a new, invisible form of labor: botsitting.
According to The Work AI Index: UK 2026, a comprehensive study of 1,500 digital workers by the Work AI Institute (a research arm of Glean Technologies), British employees are spending an average of 5.8 hours per week essentially babysitting their AI tools. This process, termed “botsitting,” involves the tedious cycle of spoon-feeding context into LLMs, correcting hallucinations, and scrubbing outputs to make them professionally viable.
The statistics reveal a jarring disconnect between adoption and actual efficiency. While 90 percent of the surveyed workforce is now required to use AI in their daily roles, and a staggering 39 percent utilize four or more different tools weekly, the perceived impact on the bottom line is marginal. Only 18 percent of respondents agree that AI has significantly improved their organization’s overall performance.
The Integration Layer is Human
The report suggests that the time AI “saves” is not being redirected toward high-value strategic work, but is instead being absorbed by the operational friction of the tools themselves. In a revealing trend, the data indicates that for every hour spent generating an output, employees spend roughly another hour refining it.
This inefficiency is driven by a high failure rate. Approximately 36 percent of AI sessions fail entirely, forcing workers to restart the process or engage in substantial reworking. This often manifests as “context loading”—the repetitive act of uploading documents, defining parameters, and reminding the AI of constraints that should, in a mature ecosystem, be persistent.
The technical friction persists despite the rollout of frameworks like the Model Context Protocol (MCP) and various API integrations. While these standards allow tools to communicate, the Work AI Institute notes they fail to solve the “context problem”: the nuanced understanding of which version of a document is current or which specific business logic applies to a given task. Consequently, the human worker has become the de facto integration layer, manually bridging the gap between the AI’s raw output and the company’s actual needs.
The Fatigue Factor and Regulatory Risks
The long-term danger of botsitting isn’t just lost time; it is the erosion of quality control. The study finds that workers are experiencing “AI fatigue,” leading 70 percent of users to eventually cut corners by passing on the first output that looks “good enough” without rigorous verification.
This lack of diligence is particularly concerning given where AI is being deployed. The report highlights that AI has moved beyond simple content drafting and into high-stakes HR functions. More than half of UK workers are comfortable with AI influencing performance evaluations, and nearly 40 percent state it is already integrated into their reviews.
Interestingly, the UK shows a higher tolerance for AI in hiring, promotion, and compensation decisions compared to the US. However, the report notes that UK firms are more cautious regarding AI-driven terminations, primarily because the UK’s more stringent employment laws make such decisions harder to defend in court than in the US’s “at-will” employment environment.
From Adoption to Transformation
The overarching conclusion from Dr. Rebecca Hinds, head of the Work AI Institute at Glean, is that the UK has prioritized the speed of adoption over the depth of operational integration. By treating AI as a plug-and-play solution rather than a workflow that requires redesign, companies have inadvertently created a new layer of administrative overhead.
“Adoption alone doesn’t equal transformation,” Hinds stated. “If employees are spending the productivity dividend on botsitting, companies haven’t eliminated work – they’ve created a new layer of overhead.” For the UK to turn its high adoption rates into a competitive advantage, the focus must shift from how many tools are used to whether the resulting work is objectively better, rather than just faster to produce.
