When people learn an AI was involved, do they just read harder, or does what they already believe about AI drive the doubt?
Can audience attention alone explain why disclosure triggers stronger skepticism?
This explores whether the skepticism people show when told content came from an AI is simply a matter of paying closer attention, or whether other forces are at work, such as prior beliefs about AI, expectations about what should be disclosed, and the absence of useful evidence.
This explores whether disclosure makes people more skeptical just because it makes them look harder, or whether something else is going on. No study in the collection pits attention against other explanations head to head, but taken together the notes suggest attention is only part of the story, and probably not the most interesting part.
Start with what attention seems to explain. When audiences learn that an AI was involved, they do become more critical and scrutinizing. Yet 34–62% of them stay persuaded anyway Does telling people an AI wrote something actually stop them from believing it?. If skepticism came only from closer reading, more scrutiny should cut persuasion much more than that. One reason it doesn't: some persuasion works below the level that scrutiny reaches. Presuppositions, meaning claims slipped in as already-accepted background ("even experts now agree..."), persuade better than direct assertions precisely because they never get flagged for evaluation Why are presuppositions more persuasive than direct assertions?. Heightened attention aimed at the wrong layer of the text leaves this untouched.
The second problem is that attention needs something to work with. In one study, readers with no provenance cues couldn't tell fabricated content from true content at all. Their discernment came back only when an interface showed which claims had been verified Can readers tell truth from fabrication without evidence signals?. Looking harder doesn't help much when there's nothing to check. Scrutiny can also lock onto misleading cues. Users trust answers with more citations even when the citations are irrelevant Do users trust citations more when there are simply more of them?. So 'paying attention' can mean leaning on shortcuts rather than evaluating the content.
The more revealing evidence points to what people bring with them rather than how carefully they read. In debate data, readers' political and religious views predict who wins better than anything about the debaters' language Does what readers believe matter more than what debaters say?. Applied to disclosure, an 'AI-written' label probably triggers beliefs people already hold about AI, not just a closer reading. The timing pattern supports this. People initially avoid AI partners once identity is revealed, but that bias reverses after repeated interactions where they can see the results Does revealing AI identity help or hurt user trust?. A pure attention effect wouldn't fade with experience. A prejudgment that gets corrected by evidence would. Norms play a part too. Readers think disclosure is more necessary than writers do, especially when the AI text is irreplaceable Do readers and writers differ on AI disclosure necessity?. That suggests some skepticism is a reaction to a perceived social contract about who did the work.
What you might not expect: disclosure also changes the social frame, not just the reading mode. People share differently with machines because no human is judging them How do people decide what to share with AI systems?, and a single strong cue like a voice can make an AI register as a social actor Do more social cues always make AI feel more present?. Labeling something 'AI' may move it out of the 'person I extend good faith to' category, and that is a different mechanism from attention. The likely answer is that attention explains the scrutiny, but prior beliefs, disclosure norms, and missing evidence explain why that scrutiny is uneven, short-lived, and only partly protective.
Sources 9 notes
Audiences aware of AI involvement became more critical and scrutinizing, yet 34–62% across groups remained persuaded. Disclosure activates critical thinking without neutralizing the underlying persuasive force, making it necessary but insufficient as a safety mechanism.
Experimental evidence shows presuppositions with additive, iterative, and factive triggers persuade audiences more than assertions, especially for discourse-new content. The mechanism: presuppositions bypass evaluative scrutiny by presenting claims as already-accepted background.
In an 81-person study, participants given no provenance cues showed no significant truth discernment (p = .43), falling for fluent hallucinations as readily as ground truth. An idealized Provenance Density interface showing verified claims restored a +4.15 point gap (p < .001).
Analysis of 24,000 Search Arena interactions shows irrelevant citations boost user preference (β=0.273) nearly as much as relevant citations (β=0.285), indicating citation count functions as a decoupled trust heuristic.
Analysis of debate corpora shows that political and religious ideology labels of voters outpredict linguistic features when modeling debate outcomes. Language effects observed without reader controls are confounded by audience composition correlated with debate topics.
Show all 9 sources
Users initially avoid AI partners when identity is revealed, but this preference reverses after repeated interactions with visible results. The learning mechanism—observing consistent outcomes—is essential; disclosure without feedback produces no calibration.
A 727-person vignette study found readers consistently rated AI disclosure as more necessary than writers did. Disclosure seemed most necessary when AI text was directly incorporated and irreplaceable, while writer effort had no effect on these judgments.
Conversational AI creates a paradoxical disclosure environment where the lack of human judgment simultaneously facilitates intimate self-disclosure (users reciprocate emotional sharing) and incentivizes deception (people self-select toward machines to avoid the psychological cost of lying to humans).
Research shows individual primary cues like voice or appearance are sufficient to evoke social-actor presence, while multiple secondary cues cannot. Quality of cues matters more than quantity in driving social responses.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- Humans learn to prefer trustworthy AI over human partners
- Exploring the Role of Prior Beliefs for Argument Persuasion
- Understanding Reader Perception Shifts upon Disclosure of AI Authorship
- Penalizing Transparency? How AI Disclosure and Author Demographics Shape Human and AI Judgments About Writing
- Toward Meaningful Transparency for AI Chatbots: Disclosing Persuasive Intent Reduces Persuasion
- Seeing to Think? How Source Transparency Design Shapes Interactive Information Seeking and Evaluation in Conversational AI
- What Influences Readers' and Writers' Perceived Necessity of AI Disclosure?
- Presuppositions are more persuasive than assertions if addressees accommodate them: Experimental evidence for philosophical reasoning