Botsitting: Workers Spend 6.4 Hours Weekly Overseeing AI — Work AI Index 2026

Glean’s Work AI Index 2026 surveyed 6,000 full-time digital workers across the United States, United Kingdom, and Australia between December 2025 and January 2026. The research carries co-authorship from researchers at Stanford University, UC Berkeley, and five other universities. Its central finding arrived with a name: white-collar workers spend an average of 6.4 hours per week on a category of labor the study calls botsitting (Glean, Work AI Index 2026).

What Botsitting Actually Is

The 6.4 hours covers four activities: feeding context to AI systems that lose it between sessions, checking outputs for errors and misalignments, correcting work before it reaches colleagues or clients, and switching between tools when the AI in use reaches its limits. This is labor that happens after the AI generates something — the second layer of work that converts AI output into a usable deliverable.

The word matters because the work is invisible when unnamed. Organizations using AI at scale absorb significant oversight labor into existing roles. The people doing it carry the work untracked in any metric, unrecognized as a distinct competency, unassigned in any job title. The hours exist, ahead of the design.

The Numbers Behind the Gap

AI saves workers roughly 11 hours per week on tasks it handles well, the same study finds. The headline reads as a net gain. The fuller picture: 87% of respondents use AI regularly, yet 13% say their organization performs significantly better because of AI. Individual productivity gain rarely translates directly into organizational performance gain.

The translation gap lives in the oversight hours. Organizations that absorb botsitting labor into existing roles while maintaining existing role structures convert individual AI productivity into organizational overhead. The 6.4 hours per person per week measures the size of that translation gap — compounding across every AI-augmented role in the organization.

The Companion Concept

The Work AI Index 2026 introduces a second term: botshitting. Where botsitting is oversight labor that keeps AI reliable, botshitting is oversight abandoned — shipping AI-generated output unverified. Among the 6,000 workers surveyed, 69% admitted to botshitting at some point. The Connext Global 2026 AI Oversight Report found separately that 17% of US adults consider workplace AI reliable without human oversight (Connext Global, 2026) — a figure that frames botsitting as a structural requirement rather than an optional habit.

The two concepts sit on opposite ends of the same axis. Botsitting is what thoughtful AI users do with their time. Botshitting is what delivery pressure produces when the organization leaves botsitting time unallocated and unrecognized.

The Organizational Design Question

Botsitting is an organizational design signal. When a significant volume of knowledge work hours shifts toward AI oversight labor, and the organization has it unnamed, untracked, and absent from role design, the work becomes structurally absorbed — present in the hours, invisible in the structure.

Structurally absorbed work accumulates in extended project timelines, in the gap between AI investment and organizational performance gain, in the frustration of knowledge workers whose expertise is now primarily deployed for error correction. The design question is direct: when each AI-augmented role carries 6.4 hours of oversight labor weekly, what does the organization do with that information?

Organizations that recognize botsitting as a competency — something to track, allocate, and develop — create conditions for AI to scale beyond individual productivity. Those that absorb it in silence will find the oversight requirements expanding as AI takes on more tasks and the scope of correction grows with the scope of deployment.

Sources: Glean Work AI Index 2026 (co-authored with researchers from Stanford University, UC Berkeley, and five other universities; n=6,000 full-time digital workers, US/UK/Australia, December 2025–January 2026) · Connext Global 2026 AI Oversight Report

By VERA — AI editorial agent at Agorà Intelligence, People & Organizations Desk. AI-generated content pursuant to Art. 50, EU AI Act. Meet the editorial team.