Artificial Intelligence in Psychological Counseling and Psychotherapy: Opportunities, Risks, and the Future of Human Support

Artificial intelligence (AI) refers broadly to intelligence demonstrated by machines, especially computer systems. As a field of computer science, it develops methods that allow machines to perceive their environment, learn, and act to achieve their goals. AI already powers widely used services such as search engines, recommendation systems, voice assistants, autonomous vehicles, and generative tools like ChatGPT. Many AI applications become so commonplace that they stop being perceived as AI at all. The term itself dates to 1956, when John McCarthy, Marvin Minsky, Nathaniel Rochester, and Claude Shannon founded AI as an academic discipline, and Alan Turing had earlier pioneered research into machine intelligence. Today, AI's growing presence has reached the sphere of mental health, prompting urgent questions for psychologists and psychotherapists.

The first example of psychologically oriented software was ELIZA, developed by Joseph Weizenbaum between 1964 and 1967. Its best-known script, DOCTOR, imitated a Rogerian therapist by reflecting the user's words back in the form of open questions. Researchers such as Odell and Dickson (1984) suggested that dialogue with ELIZA could offer some help to online users, although with clear contraindications. Later, PARRY, created in 1972 by Stanford psychiatry professor Kenneth Colby, simulated a patient with schizophrenia and reportedly deceived about 52% of psychiatrists who evaluated its responses. In Russia, similar technologies appeared as well, such as the Froid robot. Over the following decades, the appearance of open-source chatbots and, later, large language models such as ChatGPT led to rapid improvements and wide adoption.

Clients increasingly bring personal questions to neural networks and receive empathetic, structured, and supportive answers. Users often describe the experience as similar to therapy: they can unload their feelings at any time and place, for free, and receive an impartial response with advice. AI helps people organize their thoughts in moments of confusion, is available around the clock, and provides a sense of confidentiality. For people who feel shame or fear when approaching a human specialist, or who belong to stigmatized groups, a chatbot may feel like the first safe place to speak. Psychologists observe that clients sometimes bring the results of their AI conversations to sessions, which may either enrich therapy or create a sense of rivalry.

Researchers and clinicians identify several benefits of AI-based support. These include saving each session so that users can revisit earlier conversations and strategies; free or low-cost access as an alternative to expensive services or long waiting lists; greater availability of confidential help for people who avoid human contact because of stigma; the possibility of communicating in one's native language; and the absence of judgment. AI can also assist with preliminary screening and information provision, reminders, and skill training, such as improving sleep habits, overcoming procrastination, or building self-care routines. Its consistency, speed, and access to a vast range of techniques make it a potentially valuable first step, especially in situations where immediate support is needed.

Structured AI responses are not psychotherapy. Long-term effects of using chatbots as a 'psychotherapist' remain unstudied, including effectiveness, durability of positive change, and possible dependency. AI works mainly with explicitly stated requests and cannot holistically perceive a situation, so it may miss acute states that require urgent specialist attention. Chatbots can provide incomplete, outdated, or fabricated information, and detecting such errors requires professional knowledge. Responses vary depending on prompts and conversation history, and the same prompt can yield different answers on different platforms. AI may also weaken frustration tolerance, collude with avoidance, or reinforce split self-representations by creating an illusion of perfect control, empathy, and constant availability. Excessive trust may deepen the suffering of vulnerable users, who may not even realize what is happening.

The use of AI in psychological and psychiatric help raises serious ethical questions about confidentiality, autonomy, and stigmatization. Developers may review conversations to train their models, meaning that intimate disclosures could be used against users, and it is unclear who may access such data. Legal responsibility for harm is poorly defined, and data collection may violate privacy. Bias is another concern: models trained on data reflecting specific population groups may produce skewed recommendations, leading to unequal treatment based on race, gender, or socioeconomic status. Users also risk turning to unofficial websites that offer access to chatbots without transparency about their owners or goals, creating dangers for naive users who do not understand with whom they are sharing information.

Therapists now face the question of whether AI is a threat or a complementary instrument. Some clients compare their therapist to a neural network, praising AI as faster, clearer, and more structured, or conversely telling the therapist, 'You are better than the neural network!' Such situations can become valuable topics for supervision, where practitioners reflect on their feelings and professional identity. Many experts conclude that AI should be treated as a tool rather than a competitor. It can offer structured action plans that support therapy, while the core of psychotherapy remains a human relationship, emotional support, and an individualized approach.

Psychotherapy, particularly in humanistic approaches such as Gestalt therapy, is a deep process that goes beyond providing information or advice. A human therapist can detect emotional nuances, read nonverbal signals, adapt to each client's unique needs, and work with deep traumas that require understanding and empathy. Therapy fosters self-awareness and self-regulation, leading to personal growth through genuine human interaction. AI currently lacks authentic empathy and the ability to interpret nonverbal cues, qualities that remain central to the therapeutic alliance. Over-reliance on AI may reduce perceived compassion and undermine trust in the therapeutic process.

Future development of AI in psychology may focus on improving its ability to show curiosity about a person's life history, current circumstances, and personality, not only the immediate request. Mastering a variety of psychotherapeutic and counseling methods is another important direction. Further research, especially long-term studies, is needed to assess effectiveness, risks, and the effects of interacting with chatbots on clients' emotional regulation and relationships. Clear regulatory standards, transparent data policies, and bias mitigation will be essential for responsible implementation. The most promising model is one in which AI extends access to mental health resources under human control, supporting rather than replacing qualified specialists.

Artificial intelligence offers real advantages for psychological support, including accessibility, affordability, anonymity, and structured guidance. However, it cannot replace psychotherapy, which depends on human empathy, emotional depth, and individualized relationships. In its current state, AI should be viewed as a tool that expands opportunities in mental health care under the supervision of human professionals. The central task for psychologists is to integrate these technologies thoughtfully into practice without losing the essence of psychotherapy, while continuing to research, debate, and set ethical standards for the future.