NEW YORK — What happened
Researchers tracking knowledge worker habits have released telemetry data showing a massive surge in generated artifacts—emails, reports, and slide decks—following widespread LLM deployment in early 2026.
Why it matters
Quantity is easily measured, but corporate value relies on quality. If employees spend more time editing bloated AI outputs than doing original thinking, net productivity may actually be declining.
Deep dive
The study monitored software engineers, product managers, and legal analysts over a ninety-day period. While the volume of written documentation doubled, peer review scores for strategic clarity dropped by 18%. Furthermore, 'prompt debt' emerged as a major time sink, where workers spent hours refining prompts to fix nuanced context losses.
Report check (claims vs what is verified vs still rumor)
Software toolmakers claim a 55% boost in general office productivity. Independent time-motion studies verify a 20% speedup in routine drafting, but verify zero net time savings once revision cycles are factored in.
Open questions
Can upcoming training methodologies teach models to prioritize concise relevance over verbose completion, or is verbosity an inherent byproduct of next-token prediction?