Moderate evidence
More
Slow-typing AI replies earned more trust than instant ones in two small experiments
Study title: The effect of response typing speed on trust in AI chat interfaces
- Evidence type
- Controlled or validated comparison
- Sample
- n = 139
- Sources
- 1 sourcePeer-reviewed
- Published
- 2 min read
- Truvace record
- TRV-2026-3077cite or check it
Study title: The effect of response typing speed on trust in AI chat interfaces
Two experiments tested whether the speed at which an AI chatbot's reply appears changes how much people trust it. In Experiment 1, 71 participants talked with three AI agents using pre-scripted answers about exotic animals, with replies shown slowly, quickly or instantly. Typing animations raised trust in the AI agent compared with instant replies, but did not raise trust in individual responses.
Read the study abstract
Trust in artificial intelligence systems has become increasingly important as these systems assume advisory and decision-support roles. The present studies examine whether the temporal presentation of artificial intelligence (AI) generated responses, specifically typing animation speed, affects trust in chat-based interfaces. In Experiment 1, participants ( N = 71) interacted with three AI agents using pre-scripted question-response pairs detailing different facts about exotic animals. Response presentation spee…
The problem
In two experiments with 71 and 68 participants, AI chat responses shown with a typing animation earned higher trust ratings than instant responses, even though content was unchanged, so a display effect can shift trust.
In brief
In Experiment 2, 68 participants used a live language model and asked their own questions. Here presentation speed shifted both trust in the agent and trust in individual responses, with typing animations consistently producing higher trust than instant presentation.
The authors conclude that presentation timing can shape trust judgments regardless of what the AI says, which means a design choice rather than answer quality may sway how much people rely on AI advice. The work was done with small samples of participants in controlled tasks.
Main points
- Experiment 1 had 71 participants who interacted with three AI agents using pre-scripted question-response pairs about exotic animals.
- Response speed (slow, fast or instant) was varied within subjects, so each participant saw all conditions.
- In Experiment 1, typing animations raised trust in the AI agent compared with instant presentation but not trust in individual responses.
- Experiment 2 had 68 participants who used a live language model and wrote their own questions.
- In Experiment 2, presentation speed influenced both trust in the agent and trust in individual responses.
- The authors say timing features can shape trust independent of content quality.
The full story
The research was published in Computers in Human Behavior Reports in November 2026. It sets out to test whether the animated typing display common in chatbots influences trust, as AI systems take on advisory and decision-support roles.
Experiment 1 used pre-scripted question-response pairs about exotic animals with three AI agents, so content was held constant. Typing animations raised trust in the agent but not in individual responses compared with instant display.
Experiment 2 moved to a live language model and participant-generated questions to approximate real use. Under these conditions both trust measures were higher with typing animations than with instant replies.
Sources
1 on file- Computers in Human Behavior Reportsdoi:10.1016/j.chbr.2026.101451
Talk it over
In both experiments, typing animations earned higher trust than instant replies, though the content was unchanged. What does that suggest about how much a display choice should count when people judge an AI's advice?
Slow-typing AI replies earned more trust than instant ones in two small experimentshttps://truvace.com/item/the-effect-of-response-typing-speed-on-trust-in-ai-chat-interfaces
In both experiments, typing animations earned higher trust than instant replies, though the content was unchanged. What does that suggest about how much a display choice should count when people judge an AI's advice?
Answer in the threadExperiment 1 used pre-scripted answers about exotic animals with 71 participants; Experiment 2 used a live language model with 68 participants. How does each setting limit what we can say about everyday chatbot use?
Answer in the threadWhat result from a different topic, such as health or money questions, or from a larger group of participants would change your mind about whether typing speed shapes trust?
Answer in the thread
If you have more time
Look for a chatbot you have seen that shows typing animations. Who chose that design, and what might they gain if people trust it more?
Answer in the threadExperiment 1 found trust rose in the agent but not in individual responses, while Experiment 2 found both rose. What would you test next to explain the difference?
Answer in the thread
ace

What readers say