Glean: UK AI Botsitting Consumes 5.8 Hours Weekly
Key insights
- UK workers spend 5.8 hours weekly botsitting, consuming nearly half of AI's 12-hour-per-week productivity savings.
- Only 18% of organizations report meaningful AI performance gains despite 90% of workers being mandated to use AI tools.
- Thirty-six percent of AI sessions fail outright, and 70% of workers accept first-pass outputs without rigorous verification.
Why this matters
The 5.8 hours-per-week botsitting figure quantifies a real cost that most AI business cases do not model: the human supervision burden that scales with tool count and failure rate. For founders and technical leaders betting on AI productivity gains, the finding that only 18% of organizations see meaningful results despite 90% worker adoption rates shifts the conversation from deployment to workflow redesign. The 36% session failure rate and 70% acceptance-without-verification rate together suggest that AI tool reliability, not just raw capability, is the unsolved problem that will determine ROI.
Summary
UK workers are losing 5.8 hours weekly to 'botsitting': loading context windows, catching hallucinations, and verifying AI outputs, per the Work AI Institute.
AI automation saves roughly 12 hours weekly, but nearly half is consumed by supervising and fixing those same tools. Thirty-six percent of sessions fail outright and require a restart.
Essentially: (Glean Technologies, Work AI Institute) show mandatory AI rollouts have added overhead rather than eliminated work.
- Only 18% of organizations report significant AI performance gains, despite 90% of workers being required to use the tools.
- Seventy percent accept the first output that looks 'good enough' without thorough verification.
Dr. Rebecca Hinds: companies haven't eliminated work, they've created a new layer of overhead.
Potential risks and opportunities
Risks
- Glean Technologies faces credibility scrutiny since the Work AI Institute is its own research arm, making findings on poor AI ROI potentially self-serving for Glean's own enterprise productivity platform.
- UK employers mandating AI use despite only 18% reporting meaningful gains risk internal resistance and attrition pressure, particularly as nearly 40% of workers report AI is already factored into performance reviews.
- The 70% rate of unverified output acceptance creates compounding accuracy risk in regulated sectors where unchecked AI errors carry legal and compliance liability.
Opportunities
- Enterprise AI governance and observability vendors gain a direct sales narrative: the 5.8-hour botsitting cost gives IT buyers a measurable business case for tools that track AI failure rates and automate output verification.
- AI vendors with demonstrably lower session failure rates can use the 36% benchmark for competitive displacement in UK enterprise accounts, where procurement teams now have a concrete cost figure to argue for reliability over feature count.
- Glean itself, as research funder, can directly position its enterprise search product against the context-loading component of botsitting, since the report's core finding is the cost of workers manually re-establishing AI context each session.
What we don't know yet
- Full survey methodology and sample size: the article names the Work AI Institute as the source but does not detail how participants were recruited or which sectors were represented.
- Which specific AI tools or vendors account for the 36% session failure rate in UK enterprise settings, and whether failures concentrate in particular use cases or tool categories.
- Whether the 5.8-hour botsitting burden varies meaningfully by industry sector, seniority level, or number of AI tools used concurrently.
Originally reported by theregister.com
Read the original article →Original headline: UK Workers Waste Nearly Six Hours a Week 'Botsitting' — 36% of AI Sessions Fail Outright as Only 18% of Firms Report Meaningful Gains