How artificial intelligence reshapes our experience of human relationships. #
Posted August 11, 2026 [ Reviewed by Margaret Foley
](/us/docs/editorial-process)
Key points
- Many individuals are using artificial intelligence to replace needs historically met in relationships.
- AI can mimic empathy, but it cannot offer real empathy.
- Relying on AI for connection deepens isolation and erodes vital human relational skills.
Artificial intelligence (AI) has evolved rapidly over the last several years, with people growing increasingly attached to it. The various versions of AI have been described as having different personalities or personality traits (Jędrzejczak & Kobosko, 2025), with individuals developing strong attachments to certain versions. Many became particularly attached to GPT-4o, for example, with people reporting grieving and even having funerals when this version went away (Demopoulos, 2026).
Recent years have witnessed a rise in technology use that has even been described as similar to addiction. Mental health professionals have voiced concern that technology is replacing human-to-human relationships, contributing to increased isolation and loneliness, combined with deteriorating interpersonal skills (Hoffman & Hoffman, 2026; Schneider, 2019). Now, as individuals increasingly rely upon AI to write or polish emails, text messages, and letters, technology is being further inserted between human interactions. Given the accelerating pace of AI advancement and use, research and exploration of the ethics of AI are lagging.
At the 2026 American Psychological Association Annual Convention in August, AI companies were featured prominently as sponsors, and a wealth of presentations focused on AI. Diverse responses include mental health professionals scrambling to issue warnings, rapidly integrating AI into their practice, and finding opportunities in AI development. While some therapists worry about being replaced, others maintain confidence that AI will remain unable to replace the human elements of psychotherapy and recognize ways AI may contribute to a growing need for the depth and relational aspects of therapy (Hoffman, 2024).
Harlow’s Monkeys #
In the 1950s, Harry Harlow (1958; Harlow & Zimmermann, 1959) conducted his famous studies in which infant rhesus monkeys were separated from their mothers but offered two alternatives: a wire mother alternative that provided milk and a cloth mother alternative that in some studies provided milk and in others did not. Regardless of whether the cloth surrogate mother offered milk, it was preferred over the wire mother. When only the wire surrogate was present with no cloth mother, the infant monkeys still stayed in proximity to the wire version, but there was less emotional security evident in their behavior. Across the studies, the infant monkeys fared worse when there was no cloth surrogate option and when they were more isolated.
As AI and other forms of technology increasingly replace aspects of relationships that historically were met by primary attachment figures and close relationships, such as family and friends, the question can be asked as to whether the lessons from Harlow’s experiments may have renewed relevance. Harlow’s monkeys suffered when their social needs were met by technological means, such as the artificial bottles and constructed replications of attachment figures. While the monkeys survived, even grew comparatively in size and weight, they did not emotionally or socially flourish.
The Impact of AI on the Experience of Relationships #
In Ovsyannikova et al.’s (2025) study on empathy, participants often rated empathetic statements constructed by AI as more compassionate than those constructed by people. However, when they found out that these statements were constructed by AI, their experience of them changed to being less positive. A much larger study by Rubin et al. (2025) adds important nuance. While their study also found positive appraisals of AI responses that mimicked empathy, their results indicated that individuals chose to discuss certain topics with AI rather than with a human due to what they were seeking and the nature of the concerns. For example, participants tended to choose AI when curious about how AI might respond or when they were worried about how a human might respond, such as concerns about feeling judged. Additionally, when participants perceived the response as being from AI or aided by AI, the positive effects were reduced. Importantly, responses perceived to be by a human were less likely to contribute to negative emotions.
Empathetic statements are not the same as empathy (Bohart et al., 2025; Hoffman, 2020; Rubin et al., 2025). Increasingly, empathy has been reduced in the mental health field to a technique instead of an embodied experience. Yet, acting empathetic is not the same as being empathetic, and when empathy is a performance or technique instead of embodied, it often leads to feelings of hurt, anger, alienation, and objectification (Hoffman, 2020; see also Hoffman & Hoffman, 2026). While AI can mimic empathy, it cannot offer empathy. The deeper healing qualities of empathy are those that are embodied: the shared feelings, the openness to being moved by someone else, and care, concern, and love for the other person. When people experience wired empathy (i.e., the simulation of empathy by AI), it does not impact people in the same way as embodied empathy.
An important part of what bonds and sustains us in relationships is vulnerability and personal sharing. Vulnerability, however, is not easy, particularly on topics that the person experiences as shameful. While it may be easier to disclose personal thoughts, behaviors, and feelings that are shameful with an AI platform or bot, this serves to disconnect us from deeper forms of support and community in human relationships. New forms of isolation and loneliness emerge as people turn to technology for needs better filled by other people.
Conclusion #
When AI replaces important aspects of human social relationships and community, it contributes to a deepening of isolation. AI is not going away, and the ways that it is used will undoubtedly expand. As we move deeper into this age of AI, it is critical that we consider its potential impacts on essential aspects of being human and strive to preserve these precious parts of our existence.
References
Bohart, A. C., Shapiro, J. L., & Byock, G. (2025). Empathy in existential–humanistic psychotherapy. In L. Hoffman & V. Lac (Eds.), The evidence-based foundations of existential–humanistic therapy (pp. 133–156). American Psychological Association. https://doi.org/10.1037/0000446-006
Demopoulos, A. (2026, February 13). OpenAI retired its most seductive chatbot—Leaving users angry and grieving: “I can’t live like this.” The Guardian. https://www.theguardian.com/lifeandstyle/ng-interactive/2026/feb/13/ope…
Harlow, H. F. (1958). The nature of love. American Psychologist, 13(12), 673–685. https://doi.org/10.1037/h0047884 Harlow, H. F., & Zimmermann, R. R. (1959). Affectional responses in the infant monkey. Science, 130(3373), 421–432.
Hoffman, L. (2020). Culture and empathy in humanistic psychology. In L. Hoffman, H. Cleare-Hoffman, N. Granger, Jr., & D. St. John (Eds.), Humanistic approaches to multiculturalism and diversity: Perspectives on existence and difference (pp. 103–116). Routledge. https://doi.org/10.4324/9781351133357-9
Hoffman, L. (2024). The possibilities and the slippery slope of evidence-based practice in psychology: Why humanistic and existential psychologists must engage the conversation. The Humanistic Psychologist, 52(3), 318–324. https://doi.org/10.1037/hum0000347
Hoffman, L., & Hoffman, L. C. E. (2026). Artificial intelligence and being human: Critical thinking, creativity, and empathy. Journal of Humanistic Counseling. Advance online publication. https://doi.org/10.1002/johc.70047
Jędrzejczak, W. W., & Kobosko, J. (2025). Do chatbots exhibit personality traits? A comparison of ChatGPT and Gemini through self-assessment. Information, 16(7), 523. https://doi.org/10.3390/info16070523
Ovsyannikova, D., de Mello, V. O., & Inzlicht, M. (2025). Third-party evaluators perceive AI as more compassionate than expert humans. Communications Psychology, 3, Article 4. https://doi.org/10.1038/s44271-024-00182-6
Rubin, M., Li, J. Z., Zimmerman, F., Ong, D. C., Goldenberg, A., & Perry, A. (2025). Comparing the value of perceived human versus AI-generated empathy. *Nature Human Behaviour. *https://doi.org/10.1038/s41562-025-02247-w