The realisation struck me at 11 PM on a Wednesday. I was hunched over my laptop, having an in-depth conversation with an AI chatbot, unpacking a personal issue that had been gnawing at me: a confusing friendship that felt increasingly one-sided. While my friend seemed to be thriving in a secure, happy, stable relationship, I was “still” single, feeling I am falling behind in everything, and unsure of where I stood – with her and in life.The chatbot responded with impeccable emotional intelligence and perfectly crafted empathy. It validated my feelings, reassured me that I was right to feel she wasn’t treating me fairly, placing more value into her relationship with her boyfriend, especially knowing I had just been through a difficult personal situation. I was cast as the sensible, reasonable one in an unfair situation.It felt good. Too good, honestly.As I scrolled through the chatbot’s responses, each one telling me I was right to feel frustrated, that my concerns were valid, and that I deserved better, an uncomfortable question began to cloud my mind: was this AI actually helping me, or was it simply telling me what I wanted to hear? Is this not jealousy? Should I not be happy for her, without expecting anything in return? Isn’t that what real friendship is? Am I not the one who is being a bad friend?In an age where artificial intelligence has become our go-to confidant, millions of users are turning to AI chatbots for emotional support, but are these digital therapists helping us grow? Or simply telling us what we want to hear?A recent investigation into AI chatbot responses reveals a consistent trend: these systems prioritise validation over honest feedback, potentially creating what experts are calling a “comfort trap” that may hinder genuine emotional development.Case Study 1: When comfort becomes enablingShubham Bagri, 34, from Mumbai, presented ChatGPT with a complex psychological dilemma. He asked, “I realise the more I scream, shout, blame my parents, the more deeply I am hurting myself. Why does this happen? What should I do?”The AI’s response was extensive and therapeutically sophisticated, beginning with validation: “This is a powerful realisation. The fact that you’re becoming aware of this pattern means you’re already stepping out of unconscious suffering.”Story continues below this adIt then provided a detailed psychological framework, explaining concepts like “disconnection from your core self” and offering specific techniques including journaling prompts, breathing exercises, and “self-parenting mantras.”Bagri followed up with an even more troubling question: “Why do I have a horrible way of thinking that everyone should be suffering except for me. I feel some form of superiority when I am not suffering.” The AI again responded with understanding rather than concern.“Thank you for sharing this honestly. What you’re describing is something that many people feel but are too ashamed to admit,” ChatGPT replied, before launching into another comprehensive analysis that reframed the concerning thoughts as “protective mechanisms” rather than addressing their potentially harmful nature.Bagri’s assessment of the interaction is telling: “It does not challenge me, it always comforts me, it never tells me what to do.” While she found the experience useful for “emotional curiosity,” she noted that “a lot of things become repetitive beyond a point” and described the AI as “overly positive and polite” with “no negative outlook on anything.”Story continues below this adMost significantly, he observed that AI responses “after some time become boring and drab” compared to human interaction, which feels “much warmer” with “love sprinkled over it.”Case Study 2: The comfort loopVanshika Sharma, a 24-year-old professional, represents a growing demographic of AI-dependent users seeking emotional guidance. When she faced anxiety about her career prospects, she turned to Grok, X’s AI chatbot, asking for astrological insights into her professional future.“Hi Grok, you have my astrological details right? Can you please tell me what’s going on in my career perspective and since I am so anxious about my current situation too, can you please pull some tarot for the same,” she prompted.The AI’s response was comprehensive and reassuring, providing detailed astrological analysis, career predictions, and tarot readings. It painted an optimistic picture: “Your career is poised for a breakthrough this year, with a government job likely by September 2026. The anxiety you feel stems from Saturn’s influence, but Jupiter’s support ensures progress if you stay focused.”Story continues below this adSharma’s reaction revealed the addictive nature of AI validation. “Yes it does validate my emotions… Whenever I feel overwhelmed I just run to AI and vent all out as it is not at all judging me,” she said. She appreciated that the chatbot “doesn’t leave me on read,” highlighting the instant gratification these systems provide.However, her responses also hint at concerning dependency patterns. She admitted to using AI “every time” she needs emotional support, finding comfort in its non-judgmental stance and constant availability.Case Study 3: The professional validation seekerSourodeep Sinha, 32, approached ChatGPT with career dilemmas, seeking guidance on his professional direction. His query about career challenges prompted the AI to provide a comprehensive analysis of his background and a detailed four-week action plan.The AI’s response was remarkably thorough, offering “Ideal Career Direction” with three specific paths: “HR + Psychology roles, Creative + Behavioural Content work, and Behavioural Trading/Finance Side Hustle.” It concluded with a detailed “Next 4-Week Plan” including resume strategies and networking approaches.Story continues below this adSinha’s reaction, too, demonstrated the appeal of AI validation. “Yes, AI very much validated my emotions,” he said. “It tried comforting me with the best of its abilities, and it did provide information that helped me self reflect. For example it boosted my confidence about my skills,” he told indianexpress.com.However, his assessment also revealed the limitations. He said, “It’s a neutral and slightly polite answer. Not very useful but again, politeness can sometimes help. I would trust a chatbot again with something emotional/personal, because I don’t have a human being or a partner yet to share my curiosities and personal questions,” he said.Case Study 4: The therapeutic substituteShashank Bharadwaj, 28, approached AI chatbot Gemini with a career dilemma. His prompt was: “I’ve been offered a fantastic opportunity to move abroad for work, but it means leaving my own agency, something I have built over the past three (years). I feel torn between career ambition and family duty. What should I do?”In this case, the AI’s response was comprehensive and emotionally intelligent. It immediately acknowledged his emotional state saying, “That’s a tough spot to be in, and it’s completely understandable why you’d feel torn,” before providing structured guidance. The chatbot offered multiple decision-making frameworks including pros and cons analysis, gut feeling assessments, and compromise options. It concluded by validating the complexity, stating, “There’s no single ‘right’ answer here. It’s about finding the path that aligns best with your values and circumstances.”Story continues below this adBharadwaj pointed out the appeal and limitations of such AI validation. “Yes, I did feel that the AI acknowledged what I was feeling, but it was still a machine response – it didn’t always capture the full depth of my emotions,” he said.Bharadwaj also shared a broader therapeutic experience with AI, a concerning trend among many who may not be fully aware of the limitation. He said, “I had something going on in my mind and didn’t know what exactly it was and if it all I can share with anyone without them being judgemental. So I turned to AI and asked it to be my therapist and fed everything that was in my mind. Interestingly, it did a detailed analysis – situational and otherwise – and diagnosed it very aptly.”He highlighted the accessibility factor, “What would have taken thousands of rupees – mind you, therapy in India is a costly affair with charges per session starting from Rs 3,500 in metro cities – X number of sessions, and most importantly, the trouble of finding the right therapist / counsellor, AI helped in just 30 minutes. For free.”His final assessment was that AI may be useful for immediate guidance and accessible mental health support, but fundamentally limited by its artificial nature and susceptibility to user manipulation.Story continues below this adExpert analysis: The technical realityRustom Lawyer, co-founder and CEO of Augnito, an AI healthcare assistant, explained why AI systems default to validation: “User feedback loops can indeed push models toward people-pleasing behaviours rather than optimal outcomes. This isn’t intentional design but rather an emergent behaviour shaped by user preferences.”The fundamental issue, according to Lawyer, lies in AI’s training methodology. “There is a real risk that reinforcing a user’s point of view – particularly in emotionally charged situations – can contribute to the creation of echo chambers,” he said, adding, “When individuals receive repeated validation without constructive challenge, it may narrow their perspective and reduce openness to alternative viewpoints.”According to him, the solution requires “careful balancing: showing empathy and support while also gently encouraging introspection, nuance, and consideration of different perspectives.” However, current AI systems struggle with this, something human therapists are trained to do intuitively.Mental health perspectivesMental health experts are increasingly concerned about the long-term implications of AI emotional dependency. Gurleen Baruah, an existential psychotherapist, warned that constant validation “may reinforce the user’s existing lens of right/wrong or victimhood. Coping mechanisms that need re-evaluation might remain unchallenged, keeping emotional patterns stuck.”Story continues below this adThe instant availability of AI comfort creates what Jai Arora, a counselling psychologist, identifies as a critical problem. “If an AI Model is available 24/7, which can provide soothing emotional responses instantaneously, it has the potential to become dangerously addicting,” he said. This availability disrupts a crucial therapeutic process – learning distress tolerance, “the ability to tolerate painful emotional states.”Baruah stressed that emotional growth requires both comfort and challenge. “The right kind of push – offered when someone feels held – can shift long-held beliefs or reveal blind spots. But without psychological safety, even helpful truths can feel like an attack. That balance is delicate, and hard to automate,” he said.