Does a chatbot's charm fade with familiarity, the way first impressions of a new acquaintance often do?

This explores how people's views of a chatbot change over many conversations, not just after a first impression. The main thing the corpus shows is that a first impression is a poor guide to what comes later. Longitudinal studies with the Mitsuku chatbot found that the social pull that makes people warm to a chatbot early on fades in a fairly predictable way as the novelty wears off. So a one-session study can't tell you how a chatbot will be received after a month of use Do chatbot relationships lose their appeal as novelty wears off?. One framework even treats 'how many times you've talked to it' as its own variable, alongside the user's role, social setting, and the cultural stories they've absorbed about AI Why do users respond so differently to AI chatbots?.

The less obvious finding is that repeated use doesn't just wear down enthusiasm. It also raises expectations. Research on personalization shows that a chatbot that remembers you and adapts to you gains trust and feels more human over time. Each good interaction resets the baseline, though, so a later failure hurts more than the same failure would have early on. Privacy worries grow in parallel: the more the bot seems to know you, the more you wonder what it's keeping Does chatbot personalization build trust or expose privacy risks?. Trust built this way often has little to do with whether the chatbot is right. Focus groups on ChatGPT found that people trusted it for being responsive, fast, and conversational, not for being accurate Does conversational style actually make AI more trustworthy?.

Repeated interaction also changes how much people reveal. Users follow human reciprocity norms: when a chatbot consistently shares 'feelings,' people disclose more about themselves in return. Consistency worked better than trying to mirror the user Do chatbots trigger human reciprocity norms around self-disclosure?. Because nobody is judging them, people will share things they wouldn't tell another person, and the benefit seems to come from their own processing as they put it into words, not from the bot understanding them Do chatbots help people disclose more intimate secrets?. Companionship behaviors are a different story. When outside raters watched chatbots acting like companions, they found them less likable and less trustworthy Do chatbot companionship behaviors actually increase how much people like them?. What feels warm from inside a relationship can look off-putting from outside.

The most surprising effect runs the other way: talking with a chatbot can change how users see themselves. In one experiment, people's self-descriptions drifted toward GPT-4o's measured personality after personal conversations, and longer conversations produced more drift. Users ended up more alike, as if many people were being nudged toward the same template Does talking with AI reshape how people see themselves?. The chatbot changes over a conversation too. Models drift away from their original instructions within about eight turns Why do chatbots forget their instructions after a few turns?, so the 'same' chatbot a user keeps returning to may not stay the same.

The corpus is thinner on what happens when people learn how the bot works. Revealing an AI companion's strategies made some users feel more in control and others feel manipulated Does revealing an AI's strategies help or harm user agency?. Telling people a bot is trying to persuade them cuts its persuasive effect roughly in half, while just labeling it as AI does nothing, because people had usually guessed already Does telling people they are talking to AI change how persuaded they become?. There's little long-term data tracking these shifts across weeks of use. That gap is worth keeping in mind when you read claims about long-term 'devotion' to chatbots, which still rest mostly on anecdotes Are AI chatbots becoming objects of cult-like devotion?.