AI boosts worker productivity by roughly 11 hours per week, but employees spend over six hours weekly 'botsitting'—checking outputs and fixing mistakes—offsetting much of the gain.

What happened: A survey of 6,000 digital workers across the United States, the United Kingdom, and Australia (conducted between December and January) found that AI use saves each worker roughly 11 hours per week, but workers spend more than six hours per week on average checking AI output, fixing mistakes, and rerunning prompts. For every hour spent getting useful output from AI, workers spend roughly another hour making it usable. The survey also found that more than a third of AI sessions fail outright, requiring a full restart or substantial rework.

Why it matters: While 75% of individuals reported a boost in productivity, only 13% of organizations say they have seen significant business gains from AI adoption. The research suggests companies are not translating individual productivity gains into revenue and business growth. A core reason is the hidden labor cost: 37% of time workers spend interacting with AI goes to botsitting, while only 36% goes to actually producing work. The report notes there is a 'thick, mostly invisible layer of human labor holding the whole thing together.'

What to watch: The survey found 41% of workers say they sometimes deliver AI-generated work they could not explain if asked—a sign that as workers offload bigger parts of their jobs to AI, they may be offloading personal judgment and responsibility as well. The report highlights an example where a junior engineer pasted thousands of lines of AI-generated code before bed, and a senior engineer behind on a deadline had to untangle the broken code while the junior engineer struggled to explain it.