{"id":67360,"date":"2026-06-09T11:22:07","date_gmt":"2026-06-09T11:22:07","guid":{"rendered":"https:\/\/www.europesays.com\/ai\/67360\/"},"modified":"2026-06-09T11:22:07","modified_gmt":"2026-06-09T11:22:07","slug":"can-ai-ever-be-a-good-couples-therapist","status":"publish","type":"post","link":"https:\/\/www.europesays.com\/ai\/67360\/","title":{"rendered":"Can AI ever be a good couples therapist?"},"content":{"rendered":"<p class=\"duet--article--dangerously-set-cms-markup duet--article--standard-paragraph _10sa9omi itxwjq1 itxwjq0 pn04iq1\">Nick Sadler and his wife had different ideas of what a chill Saturday looked like. He considered the weekend a blank slate \u2014 no set plans, the family\u2019s moment to reset and chill. She was under the impression that time was up for grabs and put a short hangout on their calendar, which Sadler saw as his wife not taking his schedule into account. To settle the argument, he opened up ChatGPT, specifically the <a href=\"https:\/\/openai.com\/index\/group-chats-in-chatgpt\/\" rel=\"nofollow noopener\" target=\"_blank\">group chat function<\/a>, which allows more than one human to interact with the technology. Sadler prompted the chatbot to act as a neutral mediator and to instruct them on their next moves. Sadler tells Vox that ChatGPT acted as a trusted friend, or even a therapist, suggesting both of them consider different perspectives. It attempted to pinpoint where the conversation broke down (\u201cBoth of you then behaved logically according to your own understanding. That means this is not primarily a respect problem. It\u2019s a classification problem.\u201d) and offered guidelines for future scheduling (\u201cA simple question can prevent most of these arguments: \u2018Is this an idea, or are we locking this in?\u2019\u201d)<\/p>\n<p class=\"duet--article--dangerously-set-cms-markup duet--article--standard-paragraph _10sa9omi itxwjq1 itxwjq0 pn04iq1\">\u201cIt was like, \u2018Well, next time just consider this\u2019 and \u2018maybe try saying this\u2019 and \u2018maybe try doing that,\u2019\u201d Sadler, a film producer, says. \u201cWe got some sort of advice to follow, but ultimately we\u2019ve still got to do the work and we\u2019ve still got to actually take the actions.\u201d<\/p>\n<p class=\"duet--article--dangerously-set-cms-markup duet--article--standard-paragraph _10sa9omi itxwjq1 itxwjq0 pn04iq1\">Sadler, a 48-year-old self-proclaimed AI enthusiast, is no stranger to utilizing ChatGPT in his marriage. He\u2019s used it to uncover the weaknesses in his arguments and to craft apology texts to his wife. \u201cI put in purpose mistakes so she wouldn\u2019t think I was just using ChatGPT,\u201d he says.<\/p>\n<p class=\"duet--article--dangerously-set-cms-markup duet--article--standard-paragraph _10sa9omi itxwjq1 itxwjq0 pn04iq1\">But the pressures of parenting two young kids was kindling for their periodic annoying marital spats. Sadler and his wife considered couples counseling, but once he discovered ChatGPT could guide them through difficult conversations, they no longer felt they needed the help of a professional. One night, while sitting on the couch with his wife, Sadler launched ChatGPT and told his wife to talk to it as if it was a therapist. \u201cIn a way, it\u2019s having a therapist on tap,\u201d he says.<\/p>\n<p class=\"duet--article--dangerously-set-cms-markup duet--article--standard-paragraph _10sa9omi itxwjq1 itxwjq0 pn04iq1\">That people are turning to large language models to navigate their love lives isn\u2019t entirely surprising. Relationships have peaks and valleys and, many times, exist in an emotional <a href=\"https:\/\/www.vox.com\/life\/465847\/romantic-ambivalence-love-positivity-negativity-mixed-feelings\" rel=\"nofollow noopener\" target=\"_blank\">gray area<\/a>. Chatbots, on the other hand, are <a href=\"https:\/\/misinforeview.hks.harvard.edu\/article\/new-sources-of-inaccuracy-a-conceptual-framework-for-studying-ai-hallucinations\/\" rel=\"nofollow noopener\" target=\"_blank\">authoritative in tone<\/a> and <a href=\"https:\/\/link.springer.com\/article\/10.3758\/s13421-025-01755-4\" rel=\"nofollow noopener\" target=\"_blank\">confident, even when they\u2019re wrong<\/a>.<\/p>\n<p class=\"duet--article--dangerously-set-cms-markup duet--article--standard-paragraph _10sa9omi itxwjq1 itxwjq0 pn04iq1\">Some people are going a step beyond asking Claude to draft an apology text, and inviting AI into the most intimate moments of their lives: fights with their significant others. In other words, they are <a href=\"https:\/\/www.tiktok.com\/@brandon.young\/video\/7598606081069255950\" rel=\"nofollow noopener\" target=\"_blank\">treating technology<\/a> like an <a href=\"https:\/\/www.tiktok.com\/@247hustler\/video\/7504170438235344183\" rel=\"nofollow noopener\" target=\"_blank\">on-demand couples therapist<\/a>. The tech, which could be ambiently listening or addressed directly via voice or text, might suggest <a href=\"https:\/\/www.tiktok.com\/@life_with_kattie\/video\/7618787574399782157\" rel=\"nofollow noopener\" target=\"_blank\">someone use more \u201cI\u201d statements<\/a> or prompt couples to ask questions like <a href=\"https:\/\/www.tiktok.com\/@techsesh\/video\/7635046926286589214\" rel=\"nofollow noopener\" target=\"_blank\">\u201cWhere did you feel unsupported?\u201d<\/a><\/p>\n<p class=\"duet--article--dangerously-set-cms-markup duet--article--standard-paragraph _10sa9omi itxwjq1 itxwjq0 pn04iq1\">Research has suggested publicly available AI, like ChatGPT, is an <a href=\"https:\/\/arxiv.org\/pdf\/2410.07053\" rel=\"nofollow noopener\" target=\"_blank\">effective intermediary<\/a> in a dispute, with <a href=\"https:\/\/www.science.org\/doi\/10.1126\/science.adq2852\" rel=\"nofollow noopener\" target=\"_blank\">human subjects feeling less divided when AI was mediating<\/a>. But AI platforms <a href=\"https:\/\/scholarship.law.missouri.edu\/facpubs\/1368\/\" rel=\"nofollow noopener\" target=\"_blank\">lack the emotional intelligence<\/a> to adequately read a couple\u2019s body language and tone, understand cultural context and power dynamics, and incorporate a couple\u2019s past into the fight at hand.<\/p>\n<p class=\"duet--article--dangerously-set-cms-markup duet--article--standard-paragraph _10sa9omi itxwjq1 itxwjq0 pn04iq1\">The desire for an authoritative, always-available guide in the midst of conflict is certainly seductive, but emotional matters are best reserved for human-to-human conversation. \u201cThe answer is typically not that you need some type of content strategy on how you should approach your next steps,\u201d <a href=\"https:\/\/www.ameliagmiller.com\/\" rel=\"nofollow noopener\" target=\"_blank\">Amelia Miller<\/a>, a <a href=\"https:\/\/cyber.harvard.edu\/people\/amelia-miller\" rel=\"nofollow noopener\" target=\"_blank\">fellow at the Berkman Klein Center for Internet and Society at Harvard University<\/a>, tells Vox. \u201cBut it\u2019s much more that you need emotional support, which comes from asking other people that you care about what you should do in the situation, not asking a machine.\u201d<\/p>\n<p>Drawing from a shared reality<\/p>\n<p class=\"duet--article--dangerously-set-cms-markup duet--article--standard-paragraph _10sa9omi itxwjq1 itxwjq0 pn04iq1\">In her Bay Area therapy practice, <a href=\"https:\/\/www.storieswetelltherapy.com\/my-story\" rel=\"nofollow noopener\" target=\"_blank\">Courtney Quattrini<\/a> has seen her fair share of couples who leverage AI chatbots in their relationships, including using it as a practice conversation partner and to ghostwrite texts to their significant other. While none of her clients have let ChatGPT or Claude mediate a fight, some do bring in AI summaries of arguments from one person\u2019s perspective to their sessions with her. \u201cThey\u2019re ruminating or they\u2019re thinking about their side of the fight: What am I going to come back and say, how am I going to prove that I\u2019m right or wrong?\u201d Quattrini tells Vox. \u201cThey\u2019re summarizing the fight from their perspective, and then they\u2019ll bring in the summary and present it almost like it\u2019s objective, but of course it\u2019s not objective.\u201d<\/p>\n<p class=\"duet--article--dangerously-set-cms-markup duet--article--standard-paragraph _10sa9omi itxwjq1 itxwjq0 pn04iq1\">But much of the work in couples therapy centers on the idea that two things can be true at once, and is about getting both individuals to understand that their partner\u2019s emotional reality is important. \u201cWhen you\u2019re coming in and you want to summarize who won a fight, that really doesn\u2019t align with the work that we\u2019re actually doing,\u201d Quattrini says. Feeding AI your narrative doesn\u2019t help you see the things you could have done differently.<\/p>\n<p class=\"duet--article--dangerously-set-cms-markup duet--article--standard-paragraph _10sa9omi itxwjq1 itxwjq0 pn04iq1\">But when both people in a relationship invite AI into the discussion, leveling the playing field, the technology draws from a version of the story that may be more closely aligned with reality. A few months into dating, Khalid Tawohid and his partner discovered they\u2019d both been discussing their relationship with their respective AI chatbots. \u201cHow can we get our AIs to just talk to each other?\u201d Tawohid tells Vox.<\/p>\n<p class=\"duet--article--dangerously-set-cms-markup duet--article--standard-paragraph _10sa9omi itxwjq1 itxwjq0 pn04iq1\">Earlier this year, the 25-year-old software engineer designed a workaround where both his and his partner\u2019s Claude agents \u2014 drawing from each individual\u2019s full chat history \u2014 could facilitate difficult conversations. The app, called Bridge, claims to provide scaffolding for the discussions and package disorderly thoughts in a more coherent manner. Instead of looking to a machine to validate your point of view, the machine, ideally, would hold your hand as you attempt that same conversation with a human. \u201cThis helps your AI have a real sense of identity of who this [other] person is because it\u2019s two different AIs, one knows one person, one knows the other person, and they\u2019re both vehemently going to defend their own person,\u201d Tawohid says. \u201cBut together it gets you to a more shared sense of truth.\u201d<\/p>\n<p class=\"duet--article--dangerously-set-cms-markup duet--article--standard-paragraph _10sa9omi itxwjq1 itxwjq0 pn04iq1\">Still, Tawohid isn\u2019t convinced his AI chatbot mediation tool, Bridge, is even a good idea. He has shared Bridge with about 10 couples, all of whom have given him the feedback that they\u2019d use it again, he says, but it isn\u2019t widely available for use. Perhaps, he says, it could be a supplement to traditional couples counseling, a way to practice communication outside of the therapy room.<\/p>\n<p class=\"duet--article--dangerously-set-cms-markup duet--article--standard-paragraph _10sa9omi itxwjq1 itxwjq0 pn04iq1\">Ironically, though, Tawohid has come down on the side of mild AI skepticism. \u201cIt\u2019s a combination of a journal and a therapist and a friend, but it is also not real. It\u2019s also just a computer code,\u201d he says. When he discovered he\u2019d lost his ability to craft a sentence without help, he stopped writing with AI. Now he fears people could lose their relationships to chatbots, too.<\/p>\n<p>Gateway to introspection or outsourcing sincerity?<\/p>\n<p class=\"duet--article--dangerously-set-cms-markup duet--article--standard-paragraph _10sa9omi itxwjq1 itxwjq0 pn04iq1\">After a few months of using Bridge, Tawohid says he and his partner spend much less time talking to AI. They\u2019ve had enough machine-facilitated conversations that they better understand each other\u2019s thought patterns and triggers. Sadler, the AI-curious film producer, and his wife have similarly come to rely on AI less frequently because, he says, ChatGPT has taught them to be better communicators. \u201cIt just taught me to understand that she\u2019s got a different perspective on things. If I\u2019m not understanding where [she\u2019s] coming from, just asking questions to say, well, what do you mean? And not jumping to conclusions,\u201d he says.<\/p>\n<p class=\"duet--article--dangerously-set-cms-markup duet--article--standard-paragraph _10sa9omi itxwjq1 itxwjq0 pn04iq1\">Using AI as a therapeutic outlet can be instructive for people who aren\u2019t in the habit of introspection, says Miller, the Harvard fellow. These chatbots can, in theory, be a tool for reflecting on an argument and for rehearsing what to say next. But sometimes the language the chatbot suggests is so far out of the realm of what your partner would actually say that its assistance is counterproductive.<\/p>\n<p class=\"duet--article--dangerously-set-cms-markup duet--article--standard-paragraph _10sa9omi itxwjq1 itxwjq0 pn04iq1\">For Josh Elledge and his wife, the stupid fight began over a haircut \u2014 or lack thereof. Elledge, a 54-year-old podcast consultant, was refusing to clean up his look (\u201cI didn\u2019t like something my barber said, and so I stopped going to him,\u201d Elledge says) and his wife was not pleased. So she turned to an AI chatbot for assistance on how to break it to him. What she ended up saying to Elledge didn\u2019t land. \u201cIt just made her opinion stronger in a way that wasn\u2019t really helpful,\u201d he says. \u201cShe\u2019s conveying this stuff and I\u2019m like, wow, you really think that? And she\u2019s like, well, no, not really.\u201d He says they \u201cthankfully had the good sense\u201d to distinguish between what she believed and what was the AI.<\/p>\n<p class=\"duet--article--dangerously-set-cms-markup duet--article--standard-paragraph _10sa9omi itxwjq1 itxwjq0 pn04iq1\">Once you relinquish enough of your critical thinking to AI, you run the risk of undermining the relationship you sought to fix. Therapists are trained to identify when a fight needs to be slowed, rerouted, or ditched altogether. But because chatbots never tire of hearing about your problems, you can get caught in a loop of rumination, perpetually mulling over the same frustrations and workshopping language on how to tell your husband you hate his haircut. At that point, who are you in a relationship with \u2014 a large language model, or a human? \u201cThat was an instance where maybe this isn\u2019t a miracle process. You still have to just be really careful about not showing up as someone who you are not just simply because you defaulted to this AI being this authority in all things,\u201d Elledge says.<\/p>\n<p class=\"duet--article--dangerously-set-cms-markup duet--article--standard-paragraph _10sa9omi itxwjq1 itxwjq0 pn04iq1\">AI chatbots are <a href=\"https:\/\/www.library.hbs.edu\/working-knowledge\/how-ai-chatbots-try-to-keep-you-from-walking-away\" rel=\"nofollow noopener\" target=\"_blank\">programmed to keep you engaged<\/a>, but endless mediation and reflection isn\u2019t exactly helpful. If you feel compelled to use one to navigate a squabble, give the technology guardrails. For example, Miller has created custom prompts that don\u2019t exceed 10 or so exchanges with the AI and are meant to illuminate your own biases and shortcomings. But, ultimately, Quattrini, the therapist, says it\u2019s important to remember that true counsel comes from a human who possesses the ability to read nonverbal cues, affect, and changes in body language. \u201cRight now I think AI is a pretty dangerous mediator because it doesn\u2019t have a nervous system,\u201d she says.<\/p>\n<p class=\"duet--article--dangerously-set-cms-markup duet--article--standard-paragraph _10sa9omi itxwjq1 itxwjq0 pn04iq1\">The joy of being a person in a relationship with another person is getting through the hard parts together, even imperfectly. \u201cWe\u2019re complicated people and no one really knows everything going on in everyone\u2019s mind,\u201d Tawohid says. \u201cBut humans are awesome, truly.\u201d<\/p>\n<p><script async src=\"\/\/www.tiktok.com\/embed.js\"><\/script><\/p>\n","protected":false},"excerpt":{"rendered":"Nick Sadler and his wife had different ideas of what a chill Saturday looked like. He considered the&hellip;\n","protected":false},"author":2,"featured_media":67361,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[7],"tags":[16638,25,580,3771,20008,684,6445,157,4023,134],"class_list":["post-67360","post","type-post","status-publish","format-standard","has-post-thumbnail","category-openai","tag-advice","tag-artificial-intelligence","tag-chatgpt","tag-dating","tag-even-better","tag-innovation","tag-life","tag-openai","tag-relationships","tag-technology"],"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/67360","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=67360"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/67360\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/67361"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=67360"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=67360"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=67360"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}