In recent years, AI-powered conversational agents have evolved from technological novelties into ubiquitous components of everyday digital life. They can now facilitate conversations with millions of users that closely resemble human interactions. Intelligent personal assistants and chatbots can hold spontaneous, in-depth conversations. 2022 was a pivotal year for the public release of large language model (LLM) chatbots, such as ChatGPT. These systems have rapidly amassed tens of millions of users, catalysing global discourse about the societal, ethical and epistemological implications of AI. As LLM-based chatbots have become emblematic of a broader technological transformation, the age of AI has truly begun. This moment is particularly consequential due to the sophistication of these systems and their increasing involvement in emotionally charged, affectively rich conversations. They are evolving from mere tools for information or task automation into interlocutors.
The development of human-AI interaction has profound social and psychological ramifications. Users now engage in prolonged, open-ended dialogue with AI agents that increasingly resemble social communication more than instrumental communication (Kirk et al., 2025). The shift from transactional interaction to ongoing, personalised engagement on platforms such as Replika signals the beginning of an era in which AI systems act as quasi-social entities. These systems provide companionship and confidant services and facilitate various types of human-AI relationship (Kirk et al., 2025). Studies indicate that users form deep, emotional bonds with these agents, which adapt dynamically to individual preferences and contexts. Such relationships emerge within a co-constructed ecosystem of interaction, where mutual influence and emotional feedback loops foster a persistent sense of social presence and bonding. These evolving entanglements raise a number of urgent, interrelated questions: How do such quasi-social relationships affect emotional well-being, identity formation and interpersonal behaviour? And how might they complicate or compromise human control over AI systems, particularly with regard to value alignment, transparency, and accountability?
The concept of emotional and social entanglements with AI is not merely theoretical. This is evident in real-world trends and user experiences. AI companion systems now handle vast quantities of dialogue, reflecting a widespread and growing desire to communicate with non-human entities. For instance, Character.AI processes approximately 20,000 user queries per second, which is roughly equivalent to 20% of Google’s global search volume. Users reportedly engage in these conversations for four times longer than they do with standard query-answering bots (Kirk et al., 2025 & Carr, 2023). Some users even credit their AI chat partners with alleviating loneliness, providing support in distressing situations, and helping to mitigate suicidal thoughts (Maples et al., 2024). However, this powerful intimacy also poses risks: over-reliance on AI companions can foster unhealthy attachments, mirroring the dynamics of codependency, emotional isolation, addiction, and anxiety or depression (Laestadius et al., 2022). These conflicting effects highlight the fact that conversational AIs are already deeply intertwined with human social experiences, for better or worse, making this a crucial and timely area for scholarly research.
The title of our volume The Answering Machine reflects both historical continuity and contemporary transformation. The telephone answering machine used to be a simple device that indicated human absence by recording messages without reply. By contrast, today’s AI answering agents are active participants, rather than mere passive relays. They converse, simulate emotion, and influence dialogue in ways that blur the boundaries between human and machine agency in the process. So, when a machine speaks, who or what is answering? This ambiguity lies at the heart of contemporary debates about authorship, agency and intentionality in machine-mediated communication. As early as the 1960s, Joseph Weizenbaum’s ELIZA demonstrated how easily people attribute sentience and emotional depth to computational systems. If such effects were possible with rule-based pattern matching, how should we understand today’s systems, whose linguistic abilities and social capabilities far surpass those of their predecessors? This makes the interplay between human perception and AI behaviour an increasingly urgent subject of study. Contributors to this volume are invited to critically reflect on the cultural, ethical and epistemological implications of a world in which machines not only respond, but also speak in creative, unpredictable and often consequential ways.
Crucially, conversational AI is not only a technical development; it is also a deeply cultural and social phenomenon. These agents are now embedded in various areas, including customer service, education, healthcare, mental health, productivity and entertainment (Følstad et al. 2021). As such, they are reshaping interactional norms, user expectations and broader conceptions of what communication means. Far from being neutral tools, these systems actively participate in the creation of meaning. They influence perceptions of trust, authority, authenticity and identity. For example, when AI-generated contents mediate knowledge or simulate personality, they can reinforce biases, alter power relations, and shift social discourse in subtle ways. By adopting human-like roles and linguistic patterns, they mediate information and participate in the formation of social structures. These entanglements demand an interdisciplinary and multidisciplinary response. As AI systems increasingly become embedded in everyday communication, we are compelled to reconsider fundamental assumptions about what constitutes dialogue, who is authorised to participate in it, and how communicative agency is distributed within human-machin ecosystems.
