Survey: ownership of AI-aided work hinges on iteration, not skill

Among people describing recent AI-assisted tasks that felt like their own, 42 percent said they could 'definitely' have done the work without AI. Among people describing tasks that didn't feel like their own, that number was 63 percent. Ability wasn't the deciding factor; the collaboration process was.

That is the finding from a qualitative survey by researchers at Brown, Carnegie Mellon, Berkeley and the Artificiality Institute, posted to arXiv. They asked 64 participants (52 completed the full instrument) to describe two recent tasks completed with AI, one that felt theirs and one that did not.

'People disown work when they merely approve AI's suggestions, but retain ownership when they lead, iterate, or rewrite,' the paper reports. Owning the vision while handing off execution was fine; rubber-stamping output was not. Loss of personal voice and a lack of comprehension of the output both erode felt authorship.

One respondent on an AI-written email: 'it sounded nothing like me… AI emails sound very vanilla.' Another on inherited code: 'I really don't understand much of the codebase… I wouldn't consider this my work.'

The disclosure question comes apart from the ownership one. The authors write that willingness to disclose AI use is 'often decoupled from actual pride or ownership, and instead shaped by community norms and fear of credit erasure.' One participant: 'I'd absolutely mention that almost all of it was AI generated, otherwise I'd be taking way more credit than I deserve.' A couple of the researchers we track were circulating the paper this week.