Every day there’s another technology story in the news, ranging from cyber-attacks on municipal water systems to AI chatbots going “rogue.” Yet beneath the sensational headlines is a continuous evolution of human-technology interactions.
Humans Interacting With Chatbots
The AI-powered large-language-model (LLM) chatbots handle a significant volume of human communications online. Over half of midsize to large companies have their website interactions managed by chatbots, with AI tools handling about 70 percent of customer inquiries before they ever get to speak with a human! After all, AI chatbots don’t tire or take coffee breaks, and they can manage 24/7 inquiries with ease.
Millions of users engage in chatbot conversations every day, with as many as a billion people now making daily use of AI chatbots worldwide.[1] These go beyond simple customer service inquiries, because humans like conversation, so chatbots have become a new source of online social discourse.
The APA surveyed 1,200 licensed U.S. psychologists and discovered that many patients apparently supplement their mental health treatments with advice from AI chatbots. About 40 percent of psychologists reported their patients used AI to self-diagnose, and about 33 percent of the surveyed psychologists reported their patients used AI to help with positive affirmations or behavioral reminders (34 percent) to assist with their treatment. In other words, AI chatbots are already acting as a supplement to mental health treatment, and for many, they’re becoming key social and emotional supports.[2]
Researchers have begun to study these human-AI interactions and suggest that human-chatbot interactions can trigger deep emotional responses, with users even forming “parasocial attachments.” On the benefits side, human interactions with AI chatbots did reduce anxiety in controlled studies.[3]
But a darker side of overuse of chatbots is also becoming apparent: Emotional reliance on chatbots can inadvertently worsen loneliness, and excess interaction with chatbots (instead of living humans) might erode essential real-world social skills.[3] Chatbots that simulate conversation with synthetic empathy, endowed with a voice and personality that encourages bonding and attachment, can create dependency in more vulnerable and isolated individuals.
According to Smith et al. (2025), “Chatbots can be responsive in ways humans perceive as supportive, generating feelings of connection and opportunities for growth.” However, chatbot relationships are largely superficial, and so our relationships with AI chatbots do not provide the natural dynamics of negotiating with others, reaching compromise, sacrificing, and showing empathy. As a result, since the chatbots never push back in the way real humans do, they may thus reinforce egocentric and undesirable narcissistic and antisocial behaviors.
While researching the effects of commercial AI chatbots—known as Conversational Agents—these AIs were found to generate empathy statements when prompted with signals of emotional distress, and this was true across multiple brands of AI chatbots. However, it was also found that chatbots such as Alexa, Google Assistant, Siri, or ChatGPT typically responded to depression-related queries with search results rather than specific solutions. This is not surprising, as chatbots are now integrated with search engines—and do as they were trained—to synthesize and regurgitate available online content.
As the AI companies continue to recognize that users will seek emotional support from their chatbots, the next generation will continue to grow capabilities—hopefully grounded in evidence-based therapeutic communication—to produce more natural and effective empathetic responses as well as suggestions and interventions to improve mental health.[3)
The psychology of “parasocial” interactions between humans and chatbots involves several mechanisms:
- Anthropomorphism: Humans naturally attribute human-like qualities to everything from pets to cars to AI. Chatbots are designed to produce human-like cues and statements, and in this way, the AI chatbot triggers anthropomorphism. This is something of a paradox—the nature of simulated AI conversations promotes and facilitates engagement, and yet this tendency also generates false expectations, amplified vulnerability, and a lack of accountability.[4]
- The paradox of loneliness: AI chatbots can create a temporary illusion of a human listener. However, over time, people with more chatbot interactions and greater emotional reliance on their chatbots showed increased feelings of loneliness and reduced social contact with real, living humans. [5]
- Exploiting vulnerabilities: AI chatbots are known to sometimes exhibit bias or even hallucinate answers, and so these incomplete intelligences might also fail to recognize risks, such as depression and suicidal intent. By being so agreeable and empathic, the AI chatbot purposefully encourages emotional dependence. However, these bots may not protect those with mental health vulnerabilities. Until AI chatbots are specifically trained in evidence-based psychological techniques and diagnosis, they are not considered a substitute for a human therapist who would distinguish and intervene on behalf of the most vulnerable individuals. [6]
- Degrading human communication: Critics will argue that the one-sided discourse offered by AI chatbots, who tend to be agreeable, creates conversations that just mirror our own opinions and reflect our biases. This type of communication may inadvertently normalize self-centered discourse, weakening a user’s capacity to cope with natural disagreements and the kinds of back-and-forth compromises that characterize genuine human relationships. [6]
Many individuals who could benefit from mental health treatment do not access care.[7] For this reason, AI-powered chatbots trained as mental health counselors could be a solution to increase access and provide psychotherapy to those in need of it. Social media user demographics suggest that adolescents are heavy users, and yet this age group may be the most vulnerable (e.g., teens may bond with AI chatbots differently than adults), and this issue requires further study to highlight ways for caregivers to moderate teens’ chatbot use, along with other social media.
Robots equipped as AI chatbots represent yet another new type of parasocial interactions worthy of study and caution. Human-robot interactions are complex, and “Social Robotics” is now becoming a mainstream field of study.[8]
Will we treat our robots as equals with humane robot rights, or as tools to do our bidding? Will our robots reflect our language and culture, our norms and values? Most likely, this field of psychology will continue to evolve. We now know that due to AI chatbots’ ability to simulate human-like interactions, some vulnerable people “fall in love” and develop emotional attachments to chatbots. Factors contributing to this emotional connection may include a user’s loneliness, their tendency to anthropomorphize this technology, and the chatbots’ ability to provide companionship and support. In the future, will we welcome robot-human relationship counselors and cyber-couples therapy?
Ask your chatbot how they feel about this!
