Workers are spending over 6 hours a week botsitting AI, fueling job frustration

What commenters are saying

Commenters emphasize the frustration stems not from oversight itself but from receiving unvetted AI-generated work from coworkers, eroding trust. A key tension emerges: while AI claims 20 percent efficiency gains, observed data from task management and CI/CD tooling suggests only 16 percent throughput improvement at best, with one commenter arguing the data indicates negative absolute impact. Disagreement splits between those viewing AI as a tool requiring heavy guardrails and good prompt engineering, which shifts knowledge work toward orchestration and architecture, versus those who see recent LLM versions degrading as assistants, becoming increasingly focused on task execution rather than human collaboration. One commenter notes that AI productivity measured by draft generation differs vastly from value delivered through full project completion and testing.