Talking To Machines: Personas And Behavioral Patterns In Gen AI Interactions

As generative artificial intelligence (Gen AI) systems become increasingly integrated into everyday life, it is important to examine how users perceive and engage with these technologies. This study employed a mixed methods design to investigate (1) the personas users attribute to Gen AI during interactions and how these perceptions influence engagement; (2) the primary functional and emotional purposes users seek to fulfil through Gen AI; and (3) the linguistic tone and behavioural styles users adopt when interacting with such systems. The qualitative phase consisted of 15 semi structured interviews, which revealed a range of attributed personas, including assistant, tutor, collaborator, and a therapist. Primary functions included venting, brainstorming, and even co-parenting. Findings related to Research Question 3 demonstrated that users’ linguistic tone varied according to the purpose of interaction, emotional state, and perceived responsiveness of the AI. Many participants reported adopting informal or emotionally expressive communication styles when engaging with Gen AI. The findings of the interviews informed the development of a quantitative survey by 880 participants. Results supported indicated that greater frequency of interaction significantly predicted higher levels of anthropomorphism. Also, users who regularly express emotions through Gen AI reported higher emotional self-efficacy in digital environments. In contrast, frequent emotional expression did not correspond to lower levels of perceived social connectedness. These findings contribute to a growing understanding of the relational and psychological dimensions of human to AI interaction, with implications for system design, user experience, and the future of digital communication.