The rapid emergence of generative artificial intelligence (GenAI) since the debut of ChatGPT in late 2022 has fundamentally shifted the landscape of human-technology interaction, sparking both immense enthusiasm and rigorous debate. While initial adoption research largely focused on the functional utility and performance-related benefits of these tools, a groundbreaking study by Lasarov, Hoffmann, and Dwivedi (2026) suggests that the drivers of GenAI use are far more complex and socio-emotional than previously understood. Their research, published in Technology in Society, introduces a dual-value model that examines how functional value—the practical utility of the tool—and parasocial value—the sense of social presence and emotional connection—jointly shape a user’s intention to engage with conversational AI. By moving beyond traditional frameworks like the Technology Acceptance Model (TAM), the authors provide a more nuanced understanding of why individuals start and continue using these “artificial friends”.
At the heart of this research is the distinction between two primary value dimensions that are co-created through user interaction. Functional value is conceptualized as the practical benefits users derive from ChatGPT, such as enhanced productivity and assistance with daily tasks. This aligns with traditional views that technology is primarily a tool for efficiency. However, the study argues that the unique, human-like interactivity of conversational GenAI fosters a second dimension: parasocial value. Parasocial value refers to the relational value and “automated social presence” users perceive, leading some to develop one-sided emotional bonds with the AI that mimic human-to-human friendships. This socio-emotional return is particularly significant because conversational AI can interact in ways that seem indistinguishable from human communication, triggering social scripts in the user’s mind.
The researchers identify four multi-layered classes of background factors that shape these value perceptions: personality traits, evaluative beliefs (hopes and fears), institutional beliefs regarding developer responsibility, and dyadic trust. In terms of personality, factors such as extraversion and conscientiousness are critical; for instance, extroverted individuals may feel more comfortable with human-like interactions, thereby perceiving higher parasocial value. Conversely, the study introduces the “AI-as-solution belief,” which describes a user’s conviction that AI can solve societal problems and empower personal autonomy. This positive baseline appraisal indirectly boosts adoption intention by enhancing perceived functional and parasocial values. Interestingly, the research also finds that AI-related anxiety can paradoxically increase parasocial value among adopters, as attributing human qualities to the AI may serve as a coping mechanism to alleviate fear.
Trust is disaggregated in this model into two distinct components: trust in the truthfulness of the AI’s output and trust in privacy. Trust in truthfulness is essential because GenAI models are known to “hallucinate” or generate convincingly false information, which can diminish functional value if the user perceives the output as unreliable. Privacy-related trust is equally vital, as concerns about how developers handle sensitive data can cause users to interact less intimately with the tool, thereby reducing its relational or parasocial value. Furthermore, the study integrates the framework of Responsible Innovation, focusing on whether users believe developers practice anticipation, inclusion, reflexivity, and responsiveness. These beliefs are found to be more influential for current adopters, suggesting that lived experience with the tool makes the ethical conduct of the developer more salient to the user.
To test these theories, the authors conducted two interconnected studies. Study 1 utilized data from 638 participants to compare the motivations of “adopters” (those who have used ChatGPT at least once) and “non-adopters”. The results revealed a critical distinction: while functional value predicts adoption intention for both groups, parasocial value only influences the intention of those who have already experienced the tool. This suggests that while people may start using ChatGPT because it is useful, they often continue using it because of the social and emotional satisfaction they derive from the interaction. Study 2 followed 227 participants over a three-month period and confirmed that initial intentions were strong predictors of actual adoption behavior over time, although an “intention-action gap” still exists due to external catalysts or situational triggers.
The implications of these findings are profound for both organizations and policymakers. For organizations, the research suggests a “dual-track” strategy: to attract new users, the focus should remain on demonstrating utility and lowering effort barriers. To retain existing users, however, fostering a responsible sense of social presence and satisfaction can be a powerful mechanism. Policymakers are urged to address the “dark side” of these interactions, such as the risk of over-reliance or social isolation, by requiring developers to include disclaimers about the non-reciprocal nature of AI relationships. Additionally, the study highlights a burgeoning digital divide, as younger and more educated individuals are adopting these tools at higher rates. Education campaigns tailored to older demographics that focus on functional benefits and security features could help mitigate this gap.
Ultimately, this research portrays conversational GenAI not merely as a software tool, but as a complex socio-technical entity that reshapes how we define companionship and productivity. By demonstrating that adoption is a dynamic journey influenced by who the user is, what they fear, and how much they trust the developer, Lasarov et al. (2026) provide a comprehensive roadmap for the responsible integration of GenAI into society. As these “stochastic parrots” become increasingly integrated into our daily lives, understanding the balance between their practical utility and their role as artificial friends will be essential for navigating the ethical and social challenges of the future. This study stands as a vital contribution to the literature, urging a move toward value-based and stage-sensitive models of technology acceptance.
Reference: Lasarov, W., Hoffmann, S., ve Dwivedi, Y. K. (2026). With a little help from my artificial friend: Functional and parasocial value in ChatGPT use. Technology in Society, 85, 103220. https://doi.org/10.1016/j.techsoc.2026.103220