AI Pollution – Welcome to the Age of a Thousand Clippies (uberpub.com)

Everyday productivity apps have flooded users with intrusive, always-on AI widgets — draft responders, thread summarizers, tone-rewriters and sidebar brainstormers — that pop up before you’ve even read a message. The piece coins this “AI pollution”: a swarm of competing micro‑assistants (the modern Clippies) baked into Slack, Confluence, Gmail and dozens of other tools, often pushed by investors and product teams eager to claim “AI integration” rather than by clear user need. The technical and UX implications are concrete: current models frequently produce shallow, generic or tone‑stripped outputs that miss nuance, and the constant prompts fragment attention, increasing cognitive load and wasting time. The critique emphasizes that the problem is aggressive deployment and poor interaction design, not AI per se — offering Apple’s quieter, summonable approach as a counterexample. For product and ML teams the lesson is to optimize for discretion and optionality: measure interruption costs, prioritize summonable/attention‑aware interfaces, and evaluate helpfulness beyond feature checkboxes so AI becomes a real productivity aid instead of a pervasive, time‑taxing nuisance.