A Zety report finds 45% of US professionals more cautious about AI after encountering “workslop” — polished but unreliable AI output. Respondents flagged lowered trust (57%), reduced productivity (51%) and reputational risk (46%).
A Zety survey of US professionals found growing wariness about generative AI after exposure to so‑called “workslop” — AI outputs that look polished but lack accuracy or adequate human review. Forty‑five percent of respondents said encountering workslop has made them more cautious about using AI at work. teams now must show that AI actually saves time and delivers reliable results, not just tidy-looking outputs.
Respondents ranked the top operational risks tied to workslop: 57% cited lowered trust in AI, 51% said it reduced productivity, and 46% pointed to damage to company reputation. Those proportions put reliability, oversight and reputational risk at the center of organizations’ AI conversations, shifting attention away from novelty and toward governance and quality assurance for generative and agentic systems.
Business leaders interviewed for the report urged a rethink of how AI is integrated into workflows. Joel Hron, CTO at Thomson, recommends an “AI‑first, human‑second” approach: let AI attempt routine tasks initially so that humans can apply judgment and intuition at higher layers, rather than treating AI only as a final proofing pass. Framing AI as a first pass changes where human effort is applied and reframes oversight as a core part of productive workflows.
Ricoh Europe’s CIO, Nick Pearson, described a disciplined vetting model the company uses in its internal AI marketplace. The model evaluates tools for both business risk and potential financial return and explicitly asks whether a tool “really saves hours or days” and where those savings would occur. Pearson warned against automating outputs that produce little incremental value — for example, trivial meeting notes — and stressed that savings must be measured, not assumed.
For teams building or deploying AI, the report offers concrete steps: instrument and measure where AI yields time savings, design mandatory human review into critical paths, and prioritize tools that demonstrably add capability rather than cosmetic polish. The practical takeaway is that organizations must move from early experimentation to disciplined evaluation: only those deployments that prove measurable time or capability gains while preserving oversight will deliver sustained productivity improvements.
Sources
Replies (0)
No replies in this topic yet.
