Many employees are using AI to create 'workslop,' Stanford study says (www.theregister.com)

An ongoing Stanford Social Media Lab/BetterUp Labs survey finds AI is producing substantial “workslop” — superficially polished but low-value or hallucinated outputs — disrupting US workplaces. Roughly 40% of workers reported receiving AI-generated junk in the past month, and sifting it out costs an estimated $186 per employee per month in lost productivity (roughly the price of a ChatGPT Pro subscription). Recipients report annoyance (50%+), confusion (35%+), and reduced trust: 42% say they trust senders less and over a third view them as less creative or intelligent. The problem flows both ways—18% of workslop is sent upward to managers, while 16% originates from managers themselves—and sectors like tech and professional services are major generators. For the AI/ML community this flags an operational and human-risk problem beyond model capability: easy-to-produce but unvalidated outputs create overhead, erode workplace trust, and blunt any productivity gains from automation. The study underlines common failure modes (hallucinations, lack of actionable facts), the need for better prompt engineering, validation tooling, provenance/attribution and human-in-the-loop workflows, and training on when not to outsource thinking to models. It also aligns with other findings (no clear productivity lift from M365 Copilot trials; MIT reporting limited ROI for generative AI), suggesting that deployment practices and governance—not just model performance—will determine whether workplace AI helps or hinders.