AI transparency depends on who needs the answer
Smith and colleagues studied what people need from AI transparency. Their publisher abstract describes a survey of 400 people with a stake in AI and a user focus group. The point is practical: a disclosure label may not help someone check a source, opt out, or seek a correction. The full paper was not appraised here.
What the source reports
Genevieve Smith and colleagues examine what different people need to understand and challenge a generative AI system. The daily digest relies on the publisher abstract, which describes a survey of 400 people with a stake in AI and an end-user focus group. The abstract's 28% result refers to respondents, not to all users. The digest's practical reading is that a disclosure label alone may not give a person the source, a way out, or a human route to fix a mistake. The full paper was not appraised. This source does not prove any one product has a usable control.
Original source
- Title
- Whose Transparency is it Anyway? Understanding Generative AI Transparency Needs and Gaps Across Stakeholders
- Author
- Genevieve Smith, Sahiba Chopra, Natalia Luka, Min Kyung Lee
- Publication
- arXiv
- Date
- Tuesday, November 3, 2026