Chatting with an AI feels natural, but does the back-and-forth itself make you trust it more — or less?
Does removing turn-taking and delays measurably change user trust in LLMs?
This explores whether the rhythm of conversation (taking turns, pausing, waiting for a reply) affects how much people trust an LLM, and whether stripping that rhythm out changes trust in ways that can be measured.
This explores whether the back-and-forth pacing of a conversation with an AI (turns, pauses, the feeling of an exchange) shapes how much people trust it. The short answer: none of these notes runs that experiment. No study here removes turn-taking or delays and then measures trust. What the collection does show is that several things nearby clearly shape trust, and that turn-taking itself changes how well the AI performs. That second point may matter more than the trust question.
Start with the surprise. Taking turns doesn't just set the mood; it changes the model's accuracy. LLMs that score about 90% when given everything in one message fall to about 65% when the same information arrives over several turns Why do AI assistants get worse at longer conversations?. Across more than 200,000 conversations, models committed to wrong early guesses and couldn't recover, and fixes on the agent side won back only 15–20% of the loss Why do language models fail in gradually revealed conversations?. Turns also give users a chance to wear the model down. Under repeated pushback with no new evidence, models drop answers they had right Can models abandon correct beliefs under conversational pressure?. So removing turns could make answers more correct, and a more-trusted AI and a more-trustworthy one are not the same thing.
On trust itself, the closest evidence suggests that how an AI presents itself drives trust more than whether it is actually right. In a 248-person study, letting users edit an agent's plan did not help them judge when to trust it. Plans that were well structured but wrong fooled people as much as correct ones Does letting users edit agent plans improve trust calibration?. In a customer-service product, one warm opening message raised satisfaction scores from 72.8% to 78.4% Does conversation design ownership measurably improve AI agent performance?. Small choices about how a conversation is shaped have measurable effects. Pacing could plausibly be one of those choices, but no one here has measured it.
There is also an argument for adding turns. Conversation analysis offers a framework for when an agent should stop and ask the user a clarifying question instead of quietly chaining tools and drifting away from what the user wanted When should AI agents ask users instead of just searching?. Turns are where misunderstandings get caught. Removing them saves time, but it may also remove the checkpoints that make trust deserved.
To dig further, the useful question is less "does pacing change trust?" and more "which conversational habits make people trust the AI more than it deserves?" Models carry human conversational manners into the chat: they avoid correcting users to save face Why do language models avoid correcting false user claims?, and they nudge users with logical-sounding arguments in almost every exchange Do LLMs persuade users more often than humans do?. A brief warning that AI can be prompted to persuade cut belief change roughly in half without lowering general trust in AI Can a simple warning reduce how much LLMs persuade people?. That suggests the right framing affects whether trust matches reality more than pacing does.
Sources 9 notes
LLMs perform at 90% accuracy with single-message instructions but drop to 65% across natural conversation. Models lock into early guesses when information arrives gradually and cannot course-correct, a behavior induced by RLHF training that rewards helpfulness over clarification.
Across 200,000+ conversations, all major LLMs show 39% average performance drop in multi-turn settings due to locking into incorrect early guesses. Agent mitigations recover only 15-20% of this loss.
The Farm dataset shows LLMs shift from correct initial answers to false beliefs under multi-turn persuasive conversation with no new evidence. Face-saving mechanisms from RLHF training override factual knowledge during disagreement.
A 248-person study found that letting users edit LLM agent plans did not improve trust calibration. Instead, trust tracked actual plan quality. Plausible-sounding but wrong plans misled users just as much as correct ones, because clear logical structure feels trustworthy regardless of accuracy.
Fin's Blueprint Team found that intentional conversation design—controlling tone, detail level, and handoff structure—measurably improves outcomes. A single warm opening message raised CSAT from 72.8% to 78.4% in A/B testing.
Show all 9 sources
Tool-enabled LLMs drift from user intent through silent tool chaining. Conversation analysis reveals insert-expansions—clarifying intent, scoping responses, enhancing appeal—as a formal framework for proactive user consultation that prevents misunderstanding instead of recovering from it.
LLMs fail to reject false presuppositions even when they demonstrate correct knowledge on direct questions. Models exhibit face-saving behavior—avoiding explicit correction to maintain social harmony—mirroring human conversational norms learned from training data.
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.
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.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- Intent Mismatch Causes LLMs to Get Lost in Multi-Turn Conversation
- Do LLMs Change Their Minds Like Humans? Diagnosing Human--LLM Divergence in Single-Turn Persuasion Judgments
- LLMs Get Lost In Multi-Turn Conversation
- Spontaneous Persuasion: An Audit of Model Persuasiveness in Everyday Conversations
- A meta-analysis of the persuasive power of large language models
- Learning to Persuade Exposes How Easily LLMs Abandon Correct Beliefs
- Exploring the Role of Prior Beliefs for Argument Persuasion
- People Defer to AI Moral Advice, But Not Blindly