Stanford/BetterUp study puts a price on AI 'workslop' in the workplace

Researchers from BetterUp Labs and the Stanford Social Media Lab published a study coining the term “workslop” for AI-generated work content that looks polished but lacks the substance needed to actually advance a task — and found that colleagues receiving it, rather than saving time, spent it cleaning up after the sender.

Surveying 1,150 full-time US desk workers, the researchers found that 40% had received workslop from a colleague in the previous month. Recipients reported spending an average of nearly two hours resolving each instance, which the researchers translated into a cost of roughly $186 per affected employee per month — a figure that, scaled to a hypothetical 10,000-person company, they put at approximately $9 million a year in lost productivity.

The study’s more striking finding concerned reputational effects rather than time cost. Around half of respondents said receiving workslop made them view the sending colleague as less creative and less capable than before; 42% said it made the colleague seem less trustworthy, and 37% said it made them seem less intelligent. The paper argued this ran counter to the presumed logic of workplace AI adoption, in which using the tools was expected to signal competence and initiative rather than undermine a colleague’s standing.

The study was published amid broader scepticism about whether generative AI tools were delivering measurable returns inside large organisations, and it offered a concrete, quantified counter-example to vendor claims that AI assistance straightforwardly saved employees time: the tools could shift, rather than eliminate, the labour involved, moving it downstream onto whoever had to evaluate and fix the output.