Does seeing an AI wrote something make you trust it less — or does that only happen when it's disclosed?

This explores whether AI involvement makes people trust what they find less, with search results as the main case. The short answer from the corpus is: less than you'd expect in search, more in personal writing, and the bigger problem may be the opposite one. In an eye-tracking study of Google-style results pages, AI Overviews took over the visual attention that used to go to the top-ranked link. The share of attention on the first result fell from 31% to 9%. Even so, people rated the AI Overview as just as trustworthy as the ranked results Where do searchers look when AI Overviews appear?. So searchers don't seem to discount AI summaries. They read them first and trust them about the same.

The penalty for AI shows up when it is disclosed, and its size depends on the kind of writing. On a news article, adding an AI-assistance disclosure lowered ratings by less than 0.15 points on a 7-point scale. Human readers and LLM raters both showed this small penalty Does disclosing AI assistance make readers trust articles less?. For interpersonal writing, such as personal messages, revealing AI authorship lowered perceived trustworthiness, caring and likability much more. Readers saw it as breaking a social expectation, because AI can't really care How does revealing AI authorship change reader trust?. Informational content, which is most of what search returns, seems to be where people mind AI the least.

The less obvious finding is that trust in AI answers often isn't based on whether they're accurate. In 24,000 interactions with AI search engines, adding citations raised user preference almost as much when the citations were irrelevant as when they were relevant. The number of links worked as a trust signal on its own Do users trust citations more when there are simply more of them?. Users in every language tested followed confident-sounding AI answers even when those answers were wrong Do users worldwide trust confident AI outputs even when wrong?. People also trust ChatGPT partly because it feels like a conversation: it responds to them, it's fast and it's nicely formatted. None of that says anything about accuracy Does conversational style actually make AI more trustworthy?. So the risk in AI search is less that people will distrust it and more that they'll trust it for the wrong reasons.

Two related findings matter here. AI help also changes how readers see the person behind the text. In a study of nearly 3,000 writers, AI assistance shifted all 29 measured traits: writers came across as more confident, more extreme and more privileged than they really were Does AI writing assistance change how readers perceive the writer?. At the level of whole platforms, AI content can win attention while building up no lasting reputation for any human voice. This slowly weakens the way platforms help people figure out whom to trust Does AI content displace human influencers on social media?.

One gap to note: the corpus has no head-to-head test of AI-generated and human-written search results on the same query. The current evidence comes from attention studies, disclosure experiments and research on trust signals. Taken together, they suggest that labeling content as AI-made hurts trust only a little. What really drives trust is surface cues like citation count, confident tone and a conversational style, and AI search answers have all three.