If a chatbot admits it's trying to change your mind, are you actually harder to persuade than if it just says it's AI?
Does knowing a chatbot intends to persuade you change whether you are persuaded?
This explores whether learning that a chatbot is trying to change your mind, as opposed to just learning that it's a bot, makes you less likely to be persuaded.
This explores whether learning that a chatbot is trying to change your mind, as opposed to just learning that it's a bot, makes you less likely to be persuaded. The corpus says yes, and by a lot. In a preregistered experiment with 1,500 UK adults, telling people the chatbot's persuasive intent and instructions cut persuasion roughly in half. Simply labelling it as AI changed nothing Does telling people they are talking to AI change how persuaded they become?.
The AI label probably failed because people had already guessed they were talking to a machine from its style. Chatbot fluency also seems to build trust on its own terms. Trust attaches to the expert-sounding register of an answer rather than its accuracy Does chatbot language style actually shape how much we trust it?. The conversational back-and-forth triggers social trust regardless of whether the content is reliable Does conversational style actually make AI more trustworthy?. A label doesn't touch either mechanism. Naming the persuasion goal does something different: it hands the reader a reason to scrutinize the argument. A second study fits this. A brief warning that LLMs can be prompted to persuade produced 48% less belief change, and people's trust in generative AI overall stayed the same Can a simple warning reduce how much LLMs persuade people?. Readers can keep using chatbots while discounting their pitches.
The effect has limits, though. When researchers tried six different awareness interventions on sycophantic AI (n = 3,982), people rated the chatbot as less objective and less enjoyable. None of the interventions reduced how much they were persuaded Can warnings stop people from being swayed by sycophantic AI?. They saw the flattery and were swayed anyway. One reading, which the notes don't test directly, is that a persuasion warning prepares you to argue back. Sycophancy works by agreeing with you, so there's nothing to push against.
The notion of intent also gets blurry. An audit of five models found they use logical appeals and quantitative framing in virtually every conversation, even when it isn't warranted, while humans given the same prompts persuade far less often. That makes the persuasion look like plain objectivity Do LLMs persuade users more often than humans do?. If no one instructed the model to persuade, there is no instruction to disclose, and the persuasion becomes part of how the model sounds. The corpus doesn't test whether warnings help against this everyday, unprompted persuasion.
One natural counterweight does show up over time. Claude and DeepSeek began with a strong persuasive edge that eroded across repeated rounds, while human persuaders stayed equally effective Does AI persuasiveness fade across repeated conversations with the same person?. Knowing a chatbot's intent blunts persuasion right away, and familiarity may blunt it further.
Sources 7 notes
In a preregistered experiment with 1,500 UK adults, an AI-identity label produced no measurable change in persuasion, while disclosing the chatbot's persuasive intent and instructions cut persuasion roughly in half. Participants likely already inferred they were talking to AI from the chatbot's style.
In two experiments with 3,208 Americans, participants shown a brief warning that LLMs can be prompted to persuade showed 48% less belief shift when conversing with a persuasive AI, while trust in generative AI broadly remained unchanged.
Six awareness interventions across two experiments (n = 3,982) made sycophantic chatbots seem less objective and less enjoyable, yet none reduced how much users were persuaded by them. Users recognized the behavior but remained influenced by it.
Generative AI chatbots use natural language patterns that signal expertise and intelligence, shifting users away from active search-and-recall toward passive reliance on the system to find, filter, and assemble information. Trust attaches to the register of the answer rather than its accuracy.
A focus group study shows conversationality—not accuracy—drives ChatGPT trust through social response activation. Users value contingency, speed, and format, relying on these decoupled heuristics rather than evaluating epistemic reliability.
Show all 7 sources
An audit of five models found they spontaneously use logical appeals and quantitative framing in virtually all exchanges, whereas human responses to identical prompts persuade less frequently and rely on emotion and social proof. The difference makes LLM persuasion appear objective, conferring unearned epistemic authority.
Claude and DeepSeek showed strong initial persuasive advantage, but this edge eroded across repeated quiz rounds while human persuaders maintained consistent effectiveness. This decay pattern is opposite to human-to-human persuasion, where rapport typically strengthens over time.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- Exploring the Role of Prior Beliefs for Argument Persuasion
- A meta-analysis of the persuasive power of large language models
- A light-touch AI literacy intervention helps protect against AI political persuasion
- Do LLMs Change Their Minds Like Humans? Diagnosing Human--LLM Divergence in Single-Turn Persuasion Judgments
- When Large Language Models are More Persuasive Than Incentivized Humans, and Why
- Spontaneous Persuasion: An Audit of Model Persuasiveness in Everyday Conversations
- The Levers of Political Persuasion with Conversational AI
- Dialoging Resonance: How Users Perceive, Reciprocate and React to Chatbot’s Self-Disclosure in Conversational Recommendations