Babysitting the Machine: Glean's Rebecca Hinds on the Hidden Human Labor of AI at Work
TL;DR
Glean's Work AI Index 2026 survey of 6,000 workers reveals a stark disconnect: while 87% use AI and report saving 13 hours weekly, only 13% see their organization performing significantly better. The gap stems from "bot sitting" (6.4 hours of weekly hidden labor to manage AI) and "bot shit" (69% admit shipping unvetted AI outputs they cannot defend), which erode productivity gains and work quality.
đź“Š The Productivity Paradox 2 insights
Massive adoption meets elusive organizational ROI
While 87% of knowledge workers use AI and 73% report feeling more productive, only 13% believe their organization performs significantly better as a result of AI implementation.
Individual gains fail to translate to business outcomes
Workers report saving an average of 13 hours per week (a third of a work week), yet organizations struggle to convert these individual efficiency gains into measurable team or company-level performance improvements.
🤖 The Hidden Labor of Bot Sitting 2 insights
Half of AI time savings consumed by oversight labor
Workers spend an average of 6.4 hours per week on "bot sitting"—the untracked, tedious work of feeding context, debugging outputs, and cleaning up AI messes—which consumes roughly half of the 13 hours AI supposedly saves.
Alienation from automating meaningful tasks
When employees are forced to automate parts of their job they enjoy—such as customer service representatives supervising AI agents instead of talking to people—engagement drops and turnover increases.
⚠️ The Quality Crisis of Bot Shit 2 insights
Widespread delivery of unvetted AI outputs
A staggering 69% of workers admit to delivering "bot shit"—AI-generated work they cannot explain or defend—creating a "perpetual motion machine of AI slop" where polished but hollow deliverables circulate unchecked.
Shadow AI creates invisible accountability gaps
Beyond visible low-quality outputs, employees increasingly rely on AI tools without understanding their outputs, generating compliance risks and business "farce" where decisions lack human verification or expertise.
🏗️ Building Functional AI Culture 3 insights
Deep context integration reduces bot sitting
AI systems with rich organizational context—such as Glean's enterprise graph—dramatically reduce bot sitting by eliminating the need for manual context-feeding and debugging generic outputs.
Bottom-up activation trumps top-down mandates
Effective AI adoption requires bifurcated change management that pairs executive mandates with bottom-up "AI champions" who drive meaningful use rather than symbolic compliance driven by fear.
Monetary rewards and mission alignment are critical
Organizations should financially reward employees for effectively collaborating on AI solutions and align automation with meaningful shared missions to prevent the alienation that drives turnover.
Bottom Line
Organizations must shift from measuring AI adoption rates to measuring the quality of human-AI collaboration by integrating systems with deep organizational context, formally accounting for "bot sitting" labor, and rewarding employees for vetting AI outputs rather than just producing them.
More from Cognitive Revolution
View all
Intelligence on the Edge: Liquid AI's Ramin Hasani on the Search for Device-Native Foundation Models
Liquid AI CEO Ramin Hasani details how his company is building device-native foundation models using biologically-inspired 'liquid neural networks' that deliver robust out-of-distribution generalization with minimal computational resources, enabling sophisticated AI to run directly on edge devices rather than cloud data centers.
Fable's Back, AI Engineer Recap, & SambaNova
Anthropic's Fable model returns after a government safety review with refined defense-in-depth safeguards, coinciding with OpenAI's launch of GPT 5.6 Soul Ultra, creating a fragmented market where users must navigate significant pricing disparities and distinct capability trade-offs between frontier models.
1000 Designs a Day: Neural Concept's Thomas von Tschammer on AI-Native Engineering
Neural Concept is replacing days-long physics simulations with AI models that deliver results in minutes, enabling automotive manufacturers to explore thousands of designs daily rather than dozens annually. This shift allows engineers to focus on high-level trade-offs while agentic co-pilots handle iterative optimization across domains like aerodynamics, crash safety, and thermal management.
AI:AM #4: Cameron on Model Consciousness, Duvenaud's Gradual Disempowerment, swyx's AI-Eng Alpha
Consciousness researcher Cameron Berg demonstrates that frontier AI models score 30-45% on scientific consciousness indicators using automated theory-based evaluation, while revealing that internal "valence" representations governing welfare states can be directly steered to impact model safety and alignment behaviors.