Employees are losing nearly six hours every week correcting errors generated by artificial intelligence, a phenomenon now dubbed botsitting. While AI tools promise massive productivity gains, much of that saved time is immediately consumed by workers who must act as digital chaperones. A study of 1,500 UK collaborators highlights that the efficiency of automation is often neutralized by the need for constant oversight.
This labor-intensive process involves several repetitive stages to ensure quality control:
- Constant review of automated responses for factual inaccuracies.
- Aggressive prompt engineering and reframing to get specific results.
- Switching between different LLM models when one fails to deliver.
- Repeating full workflows until the output meets professional standards.
The term botsitting reflects the reality that technology has not yet reached a "set it and forget it" stage for complex tasks. Instead of offloading work, many professionals find themselves in a cycle of babysitting algorithms, ensuring that the AI does not hallucinate or produce substandard content that could compromise business outcomes.
