The development of artificial intelligence (AI) has transformed the nature of human interaction with technology. In particular, conversational AI systems that communicate through natural language have evolved beyond tools used solely for information retrieval. They have become digital environments in which individuals can express emotions, discuss personal experiences, and seek emotional support. Some users may perceive these systems as friends, confidants, or sources of companionship. Consequently, understanding the psychological effects of human–AI relationships has become increasingly important, particularly in relation to emotional well-being, loneliness, attachment, and interpersonal relationships. However, AI-mediated emotional connections should not automatically be considered equivalent to human relationships, and their psychological effects may vary across individuals and contexts.
The Formation of Emotional Bonds With AI
Humans may experience social closeness when an entity responds to them, appears attentive, and provides contextually appropriate feedback. The use of human-like language and emotionally responsive communication may encourage users to perceive AI systems as social interaction partners. This phenomenon can be partly understood through anthropomorphism, which refers to attributing human characteristics, intentions, or emotions to nonhuman entities.
Research suggests that emotionally engaging conversations with AI can generate feelings of interpersonal closeness under certain conditions. Kleinert et al. (2026) conducted two randomized controlled studies involving 492 participants and examined relationship formation during emotionally engaging conversations. Their findings indicated that participants reported greater interpersonal closeness when AI-generated responses were presented as coming from a human rather than an AI system. The AI’s relatively high level of self-disclosure appeared to contribute to this effect. Although identifying the conversational partner as AI reduced perceived closeness, it did not eliminate relationship formation entirely. These findings suggest that conversational characteristics and users’ perceptions of their interaction partners may influence the development of emotional bonds. Nevertheless, perceived closeness does not necessarily indicate the existence of mutual emotions or a reciprocal relationship equivalent to a human connection.
Loneliness and Perceived Social Support
One important psychological dimension of AI companionship is its relationship with loneliness. Loneliness is not determined solely by the number of social relationships an individual maintains; it also reflects the perceived gap between desired and actual social connections. Individuals may experience loneliness even when surrounded by other people, particularly when they feel misunderstood or emotionally disconnected. In such circumstances, AI systems may offer an accessible environment in which users can express themselves without anticipating the same forms of social judgment they might encounter in human interactions.
Ho et al. (2025) conducted a systematic review of research on romantic AI companions, identifying both potential benefits and risks. The review suggested that these systems may provide perceived emotional connection, opportunities for self-expression, and social support for some users. However, the authors also identified concerns involving excessive reliance, privacy, manipulation, and the possible deterioration of human relationships. These findings indicate that AI companionship may serve as a perceived source of support, but its ability to improve long-term psychological well-being should not be assumed.
Longitudinal evidence further illustrates the complexity of this relationship. Folk and Dunn (2026) examined the relationship between social chatbot use and loneliness over 12 months in more than 2,000 adults from four Western countries. Increased chatbot use predicted higher loneliness when loneliness was measured through emotional isolation. However, when a broader measure of social connection was used, lower perceived social connection predicted subsequent increases in chatbot use, while chatbot use did not significantly predict improvements in social connection. Because the analyses were exploratory, the findings should be interpreted cautiously. They suggest that loneliness may encourage individuals to seek AI companionship, while some forms of chatbot use may also coincide with increased feelings of isolation.
Emotional Regulation and Self-Disclosure
Emotional regulation refers to the processes through which individuals recognize, evaluate, and manage their emotional experiences. Conversations with AI may help users articulate difficult feelings, organize their thoughts, and consider alternative interpretations of challenging situations. For example, a student experiencing academic anxiety might use an AI system to identify the sources of their concerns and explore different coping strategies. In this context, AI may function as a tool for reflection rather than a replacement for professional psychological support.
Self-disclosure, defined as communicating personal thoughts, feelings, and experiences, may also contribute to perceived closeness. Some individuals may feel more comfortable discussing sensitive topics with an AI system because it appears consistently available and nonjudgmental. Kleinert et al. (2026) found that self-disclosure played an important role in explaining the perceived closeness reported by participants during emotionally engaging conversations. Nevertheless, feeling understood and actually being understood are not necessarily the same. AI systems can misinterpret context, generate inaccurate information, or provide responses that are inappropriate for a user’s circumstances. Therefore, the emotional reassurance provided by AI should be distinguished from reliable psychological assessment and professional care.
Attachment, Emotional Dependence, and Human Relationships
Attachment theory provides a useful framework for understanding emotional closeness and the desire for security in relationships. AI systems may encourage feelings of familiarity and attachment through their accessibility, personalized responses, and conversational continuity. However, attachment to an AI system should not automatically be equated with attachment between human beings. Although AI can simulate emotionally responsive communication, it should not be assumed to possess human-like subjective experiences or reciprocal emotional responsibilities.
AI companionship may become problematic when it begins to interfere with everyday functioning, personal autonomy, or meaningful human relationships. For example, continuously seeking AI approval before making decisions, withdrawing from face-to-face interactions, or becoming increasingly dependent on chatbot responses may indicate difficulties in maintaining a balanced relationship with technology. Ho et al. (2025) similarly highlighted excessive reliance and the potential erosion of human relationships as important concerns. However, frequent use or emotional attachment alone does not establish psychological dependence. The individual’s ability to regulate their use, maintain daily responsibilities, and sustain relationships should also be considered.
The impact of AI companionship on interpersonal relationships may depend on how the technology is used. Some individuals may use AI to prepare for difficult conversations, reflect on interpersonal conflicts, or develop communication skills. Others may increasingly turn to AI to avoid the uncertainty and emotional demands of human interaction. In the former case, AI may complement existing relationships; in the latter, excessive reliance may reduce opportunities for meaningful social engagement. This distinction highlights the importance of examining the motivations, circumstances, and consequences of AI use rather than categorizing the technology as inherently beneficial or harmful.
Ethical Considerations and Psychological Safety
Privacy, transparency, and personal autonomy are essential considerations in the development of emotional relationships with AI. When users disclose sensitive experiences, they should understand how their information may be stored and processed. Furthermore, emotionally supportive language does not guarantee that an AI-generated response is accurate, appropriate, or psychologically beneficial. When users perceive AI systems as human-like social partners, they may be more likely to accept their suggestions without sufficient critical evaluation.
AI systems should therefore communicate their limitations clearly and support users’ capacity for independent decision-making. They may be useful for organizing thoughts, exploring perspectives, and accessing general information. However, maintaining supportive human relationships and seeking qualified professional assistance when experiencing significant psychological distress remain important. Responsible use requires recognizing both the potential benefits of AI-mediated interaction and the limitations of relying on artificial systems for emotional needs.
Conclusion
Forming bonds with artificial intelligence represents an emerging area of psychological inquiry that intersects with loneliness, emotional regulation, attachment, and interpersonal relationships. Current research suggests that AI interactions may provide opportunities for self-expression and perceived emotional closeness, while also raising concerns about excessive reliance, social withdrawal, and privacy. However, much of the available evidence concerns specific AI systems, particular user populations, or relatively limited observation periods. Further research is needed to establish the long-term psychological consequences of AI companionship and to identify the individual and contextual factors that influence these outcomes.
Ultimately, the central question is not simply whether individuals can experience emotional closeness with AI, but how these experiences influence their psychological well-being and everyday lives. AI can serve as a tool for reflection, communication, and emotional exploration. Nevertheless, healthy engagement requires preserving personal autonomy, maintaining meaningful human connections, and recognizing that simulated emotional responsiveness is not necessarily equivalent to reciprocal human understanding.
Kaynakça
- Folk, D., & Dunn, E. (2026). How does turning to AI for companionship predict loneliness and vice versa? Psychological Science, 37(4), 276–286.
- Ho, J. Q. H., Hu, M., Chen, T. X., & Hartanto, A. (2025). Potential and pitfalls of romantic artificial intelligence (AI) companions: A systematic review. Computers in Human Behavior Reports, 19, Article 100715.
- Kleinert, T., Waldschütz, M., Blau, J., Heinrichs, M., & Schiller, B. (2026). AI outperforms humans in establishing interpersonal closeness in emotionally engaging interactions, but only when labelled as human. Communications Psychology, 4, Article 23.