Survey: 48% of Workers Fix AI Output They Receive
Use.AI's poll of 8,316 workplace AI users found 39% say verifying AI output takes longer than doing the task from scratch.
Nearly half of employees who use AI at work say they have received AI-assisted work that needed correction, according to a survey of 8,316 adults published by Use.AI. The research, reported by techradar.com, covered respondents in the UK, EU, US, Canada, Australia and Latin America.
The survey found that 43% said they had been forced to redo AI work that was unsuitable, while 34% admitted forwarding AI-produced or AI-assisted work without checking its contents. A further 39% said checking the AI's output took longer than doing the task from scratch.
Verification appears to be adding to workloads rather than easing them: 41% reported more reviewing and verifying of AI-produced material since AI entered the productivity process. Even so, 62% said AI made creating documents, memos and code easier.
Rules for checking AI-assisted work remain scarce. According to the findings, 58% said their workplace has no rules for verifying AI-assisted work.
Use.AI's research can be read online. The article was written by Christian Cawley.
Quick answers
How many people did Use.AI survey?
Use.AI surveyed 8,316 adults who use AI in the workplace across the UK, EU, US, Canada, Australia and Latin America.
How many workers say checking AI output takes longer?
39% of respondents said checking the AI's output took longer than doing the task from scratch.
Do most workplaces have rules for checking AI work?
No. 58% of respondents said their workplace has no rules for checking AI-assisted work.