Considerate Use of Generative AI in the Workplace (declanbright.com)

“AI Work Slop” describes a growing workplace habit where people use generative AI to produce long-form content (emails, reports, decks) and recipients then feed that output back into AI to summarise before reading. The result is an inefficient expansion–compression cycle that increases cognitive load, wastes tokens and time, and risks diluting or distorting intent as multiple model passes introduce omissions or inconsistencies. For teams relying on AI, this pattern magnifies miscommunication and reduces signal-to-noise, especially when different models or prompts are used on sender and receiver sides. Technically, the cycle raises costs (more tokens and compute), adds latency, and amplifies model variability—summaries can omit nuance or introduce hallucinations not present in the original. The practical fix is procedural: only generate long-form when necessary, and always include a top-line summary so recipients who need just the gist can read it immediately. A simple prompt addition—“Add a concise summary section at the top of this document”—ensures consistent, sender-authored TL;DRs, reduces redundant model calls, and preserves meaning across workflows.