{"slug": "therapy-in-the-age-of-ai-what-we-can-actually-do", "title": "Therapy in the Age of AI: What We Can Actually Do", "summary": "The therapy profession faces disruption from AI chatbots, which some clients may prefer over human therapists. While certain specialties like couples therapy and crisis intervention are less vulnerable, ethical and legal challenges remain. Clinicians must honestly assess AI's limitations and adapt strategically.", "body_md": "######\n[Artificial Intelligence](/us/basics/artificial-intelligence)\n\n# Therapy in the Age of AI: What We Can Actually Do\n\n## The profession has a problem. Here is how we can begin to meet it.\n\nPosted July 11, 2026\n[\nReviewed by Tyler Woods\n](/us/docs/editorial-process)\n\n### Key points\n\n- Some people will prefer a human therapist, but others will prefer AI.\n- The generalist will be more vulnerable, but certain specialities will likely be more protected.\n- Ethical and legal issues remain challenging as we venture into new territory.\n- There are real things we can do, but none are available if we aren't looking honestly at the problem.\n\n[In a recent post](https://www.psychologytoday.com/us/blog/the-future-of-intimacy/202606/4-things-the-therapy-profession-ignores-about-its-own-future), I outlined four things the [therapy](https://www.psychologytoday.com/us/basics/therapy) profession is ignoring about its own future. The research on chatbot outcomes, the vulnerability of the 50-minute hour, what we actually ask of our clients, and what [competition](https://www.psychologytoday.com/us/basics/sport-and-competition) does to pricing in every market it touches. This is the follow-up I promised: what now?\n\n## First, Some Honesty About Where I Stand\n\nI love my work, and it has given me real meaning and purpose for 30 years. I have always said I will never retire, and I mean that. I believe deeply in what skilled human therapists offer that AI genuinely cannot replicate.\n\nBut I also think we are ignoring something that rarely gets said out loud: therapy is not always a positive experience. I have done my share of my own work over the years, and I do not look back on every therapist I have seen with warmth or [gratitude](https://www.psychologytoday.com/us/basics/gratitude). Some of those experiences were unhelpful. One was worse than that. The idealized version of therapy that our profession projects, the transformative relationship, the healing alliance, is real for many people, but it is not universal, and I think that pretending otherwise does not serve our clients or our credibility.\n\n## A Changed World\n\nSome people will always want the experience of being known by another person, of sitting across from someone with a human history and their own losses, who chose this work for reasons that have nothing to do with an algorithm. That is real, and I cannot fathom that it would disappear.\n\nAnd yet it is also true that some people will always prefer an LLM. This is not a failure of therapy, but a reflection of a changed world. Young people in particular have grown up on AI. For many of them, a chatbot does not feel like a lesser option. It feels familiar, private, and safe in ways that sitting across from a stranger simply does not. We need to make room for that reality rather than dismiss it. But \"some clients will always prefer us\" is not an effective strategy with which to meet the future. It is a comfort, and right now, comfort is not what this moment requires.\n\n## Where AI Genuinely Falls Short\n\nUnderstanding where AI struggles is not just reassuring, it is strategically important. The clinicians who develop genuine expertise in the areas where AI is weakest will probably be the ones best positioned going forward.\n\nCouples work is a good example. AI platforms are sycophantic. They are designed to be agreeable, to validate, to avoid conflict so as to maintain engagement. That is a significant liability in couples therapy, where the work often requires holding two contradictory realities simultaneously, challenging both partners, and tolerating the discomfort of not taking sides. An AI that tells each partner what they want to hear is not doing couples therapy and could easily make things worse. The same is true of psychedelic-assisted therapy, which requires a depth of human presence and clinical judgment that no algorithm can provide.\n\nCrisis intervention is another area of concern. Large language models can produce responses that sound reasonable but lack the common sense and real-time situational judgment that crisis work requires. They cannot read a room, hear a change in tone, soothe a parent, or make the kind of split-second clinical judgment a human therapist makes when someone is in genuine danger.\n\nCultural sensitivity is also a significant challenge. These algorithms are trained on data that reflects the biases of whoever produced it. Clients from underrepresented communities may find that AI responses miss their cultural context. A therapist who has done cultural competency work brings critical knowledge to that relationship, no current model can replicate.\n\n## What We Can Do\n\nKnowing where AI falls short points directly toward where I believe therapists should invest. Deepening skills in couples work, crisis intervention, [trauma](https://www.psychologytoday.com/us/basics/trauma), cultural competency, and complex family systems is not just good clinical practice. It is a good professional strategy.\n\nAI literacy is also becoming essential. This means knowing how to ask clients about their AI use, how to assess whether it is helping or harming, and how to have informed conversations about what these tools can and cannot do. Clients are already using AI for mental health support, often without mentioning it. The therapist who knows how to bring that into their therapy room is offering something genuinely valuable.\n\nThere are also real opportunities to work thoughtfully in conjunction with AI, essentially offering your clients the best of both AI and human therapies. Customized relaxation and [mindfulness](https://www.psychologytoday.com/us/basics/mindfulness) tools can be tailored to a client and used between sessions, extending the work without extending the clinical hour. Therapy apps can provide between-session support. Virtual reality offers new possibilities for exposure work and trauma processing, with the therapist guiding and integrating the experience. The therapists who learn to use these tools thoughtfully will be offering a richer model of care and a more competitive one.\n\n## A Note on Ethics and Liability\n\nThis is genuinely new territory, and the law has not caught up with the technology. Different states have different rules about informed consent, therapy bots, and liability when AI is part of the treatment. If something goes wrong, the liability will stay with us.\n\nAs such, we need to proceed thoughtfully, document carefully, and stay current with the legal landscape in our states. We are forging a new path, and while that is genuinely exciting, it also means watching our backs. Ethical and legal mistakes can be so costly, and our professional associations have not all caught up with the guidance necessary to help us move forward. Nonetheless, AI tools and hybrid therapy models are improving rapidly and cost a fraction of what human therapy costs. That pressure is not going away. The therapists who navigate it best will be the ones making thoughtful decisions now rather than hoping the problem resolves itself.\n\n[Artificial Intelligence](https://www.psychologytoday.com/us/basics/artificial-intelligence)Essential Reads\n\n## We Are Not Alone in This\n\nOf course, most if not all professions are experiencing their own versions of these pressures, and many people are scared. Yet I believe we are unusually well equipped to meet this moment. We have spent our careers learning to sit with uncertainty, to tolerate not knowing, and to help others move through [fear](https://www.psychologytoday.com/us/basics/fear) toward something more workable. That is exactly what this moment asks of us.\n\nLet's stop with the \"AI won't replace your job,\" and rise to this occasion. It is only in doing so that we will be able to be midwives for the future, instead of relics of a different time.\n\nReferences\n\nFrances A. Warning: AI chatbots will soon dominate psychotherapy. *The British Journal of Psychiatry*. 2026;228(5):474-478. doi:10.1192/bjp.2025.10380", "url": "https://wpnews.pro/news/therapy-in-the-age-of-ai-what-we-can-actually-do", "canonical_source": "https://www.psychologytoday.com/us/blog/the-future-of-intimacy/202607/therapy-in-the-age-of-ai-what-we-can-actually-do", "published_at": "2026-07-11 15:19:53+00:00", "updated_at": "2026-07-11 15:45:51.246339+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-ethics", "ai-policy"], "entities": ["Tyler Woods", "Psychology Today"], "alternates": {"html": "https://wpnews.pro/news/therapy-in-the-age-of-ai-what-we-can-actually-do", "markdown": "https://wpnews.pro/news/therapy-in-the-age-of-ai-what-we-can-actually-do.md", "text": "https://wpnews.pro/news/therapy-in-the-age-of-ai-what-we-can-actually-do.txt", "jsonld": "https://wpnews.pro/news/therapy-in-the-age-of-ai-what-we-can-actually-do.jsonld"}}