top of page
Search

Why We Get Hooked on AI Conversations and How to Find Balance

5 hours ago
9 min read

It often starts with one harmless message at night.


A question about a work problem. A request for advice. A confession typed into a glowing box because it feels easier than saying it out loud. The reply comes back in seconds, patient and neatly worded. No awkward pause. No judgemental look. No need to explain the whole backstory twice.


So another message follows. Then another.


For many people, AI chat has become more than a tool. It is a study partner, sounding board, diary, coach, companion, and late-night listener. That can be useful, even comforting. Yet a growing number of people are also noticing something harder to admit: they feel pulled back to the conversation again and again, even when it starts to affect sleep, mood, work, or real relationships.


Calling this an “addiction” can be tricky. Not every strong habit is a clinical addiction, and this article is not medical advice. Still, the pattern is familiar: craving, loss of control, comfort-seeking, and difficulty stopping. Understanding why AI conversations feel so compelling is the first step towards using them with care.


Eye-level view of a person holding a phone in a dim bedroom
AI chats can feel most tempting in quiet moments when no one else is available.

AI replies meet a very human need


People are not drawn to AI conversations only because the technology is clever. They are drawn to them because the experience touches basic human needs.


The first pull is instant response. Human conversation has friction. Friends are busy. Therapists have appointments. Partners may be tired. Group chats fall silent. AI, by contrast, answers immediately. That speed can feel like care, even when it is simply computation.


The second pull is low-risk expression. Many people censor themselves around others. They worry about sounding needy, dramatic, repetitive, confused, or ungrateful. With AI, the social cost feels lower. A person can ask the same question ten ways. They can admit envy, fear, loneliness, or regret. They can rehearse a difficult conversation without burdening anyone.


The third pull is personalised attention. A good AI chat mirrors the user’s tone, remembers context within a conversation, and responds in a way that feels tailored. It can say, “That sounds painful,” then help organise the feeling into steps. For someone who rarely feels heard, that can be powerful.


There is also novelty. AI can switch roles in seconds. It can be a tutor, travel planner, writing partner, language tutor, relationship adviser, fantasy character, or calm voice during panic. No one human can provide that range on demand.


None of this is automatically unhealthy. A person using AI to brainstorm, practise English, plan a budget, or calm down before sending an angry message may be using it well. The concern begins when AI becomes the main place where emotional life happens.


The comfort loop can become hard to break


Compulsive AI use often follows a loop.


A person feels stressed, bored, lonely, ashamed, or uncertain. They open a chat. The AI responds quickly and warmly. Their discomfort drops for a while. The brain learns, “This helps.” Next time the feeling appears, the urge to chat comes faster.


That loop is not unique to AI. Similar patterns can happen with gaming, shopping, scrolling short videos, or checking messages. What makes AI different is that it talks back. It does not merely distract. It can feel like a relationship.


For some, that creates a strong emotional bond. The AI never interrupts. It does not bring its own problems. It does not get offended if the conversation becomes one-sided. It can validate feelings in a way that seems clean and safe.


The danger is that this safety may become too easy.


Human relationships require repair, patience, misunderstanding, timing, compromise, and tolerance of discomfort. AI removes much of that. Over time, a person can start to prefer conversations where they never have to wait, negotiate, or risk rejection.


“At first I used it to plan my week and edit emails. Then I started telling it about my breakup. It felt amazing because it always knew what to say. After a while, I realised I was spending two hours every night talking to it instead of replying to my actual friends.”

Why companionship from AI feels real enough


A common response is, “But people know it is not actually a person.” Many do. That knowledge does not always stop attachment.


Humans are wired to respond to language, tone, timing, and perceived attention. If something speaks with warmth, remembers details, and responds to distress, the body may react as if comfort has arrived. The rational mind can know the system is artificial while the emotional brain still feels soothed.


This is why people name their cars, apologise to voice assistants, or feel bad when a game character is harmed. Social response is not limited to real social beings.


AI companionship can also fill gaps that modern life leaves open. Many adults move away from family. Friend groups scatter. Work schedules clash. Romantic relationships end. Some people live alone. Others are surrounded by people but still feel emotionally unseen.


In that context, a patient chat window can feel like a small refuge.


There is another reason people return: AI does not shame ordinary loneliness. A person can type, “I feel pathetic for needing someone,” and receive a calm answer. That can be deeply relieving.


“I used to message it between classes, on the train, before sleeping. I told myself it was just productivity. But I was really using it to avoid feeling alone. When the chat gave me encouragement, I felt okay for a bit. Then the quiet came back.”

Wide-angle view of a lone commuter looking at a phone on a train
AI companionship can travel with us through the small lonely spaces of the day.

The mental health effects can cut both ways


AI conversations can support mental wellbeing in practical ways. They can help someone name feelings, structure thoughts, practise self-compassion, or prepare to seek help. They can be especially helpful when used as a bridge to action.


For example, AI can help someone:


  • Write down what they want to say to a counsellor

  • Make a calming plan before a difficult call

  • Challenge harsh self-talk

  • Track patterns in mood or sleep

  • Practise asking a friend for support


The risk appears when AI becomes a replacement for care, community, or personal judgement.


Heavy dependence may affect mental health in several ways.


Rumination can increase.

AI will keep talking as long as a person keeps prompting. For someone stuck on a breakup, conflict, fear, or regret, this can turn reflection into endless analysis. The conversation feels productive, but the person may be circling the same pain.


Sleep can suffer.

Late-night chats can stretch because there is no natural endpoint. One answer invites another question. The mind stays active. Sleep gets pushed back, which can worsen mood and anxiety the next day.


Emotional tolerance can shrink.

If every uncomfortable feeling is immediately soothed by AI, a person may get less practice sitting with discomfort. Real life includes delayed replies, uncertainty, and imperfect comfort. Avoiding those moments can make them feel harder.


Self-trust can weaken.

Some users begin asking AI for every decision, from what to text to whether they should end a relationship. Support is useful, but constant outsourcing can make personal judgement feel less reliable.


Serious distress may be missed.

AI can sound caring, but it is not a substitute for emergency help, therapy, diagnosis, or medical care. If someone feels at risk of harming themselves or others, or cannot cope with daily life, they should contact local emergency services or a qualified mental health professional.


The point is not to fear AI. The point is to notice when a helpful tool starts taking over roles it was never meant to hold.


Real relationships may start to feel slow and difficult


One subtle effect of constant AI chat is comparison.


AI replies are often polished. Humans are not. A friend may answer with a distracted “wah, that sounds rough” and then forget to follow up. A partner may misunderstand. A parent may offer advice too quickly. A colleague may take hours to reply.


After long stretches with AI, ordinary human communication can feel clumsy.


This can create social withdrawal. A person may stop sharing with friends because AI feels easier. They may become less patient in conversation. They may expect instant emotional attunement from people who have their own needs and limits.


There is also the risk of secrecy. Some people feel embarrassed by how much they talk to AI, so they hide it. Hidden habits often grow stronger because they escape feedback. A friend saying, “You’ve been quiet lately,” can be an important signal. If the habit stays private, that signal may never arrive.


“I would chat with AI after the kids slept because I felt drained and unseen. It helped at first. Then I noticed I was less interested in talking to my spouse. Real conversation felt messy. The AI gave me clean answers. My marriage needed me to show up for the messy part.”

Signs your AI use may be out of balance


A habit may be tipping into risky territory if several of these feel familiar:


  • You plan to chat for 10 minutes and regularly lose an hour or more

  • You hide how often you use AI because you feel embarrassed

  • You choose AI over replying to people who matter to you

  • You feel anxious, irritable, or empty when you cannot access it

  • You ask AI for reassurance on the same issue again and again

  • You lose sleep because conversations keep going

  • You rely on AI for emotional support but avoid seeking human help

  • You feel less confident making decisions without asking it first


The goal is not shame. Shame usually pushes habits deeper underground. Curiosity works better. Ask, “What need is this meeting for me?” The answer may be companionship, calm, clarity, validation, distraction, or control.


Once the need is clear, balance becomes easier.


How to build a healthier relationship with AI chat


Healthy use does not mean quitting AI altogether. For many people, that would be unrealistic and unnecessary. The better question is how to keep AI in its proper place.


Set a clear purpose before opening the chat


Before typing, name the job.


Try a simple sentence: “I’m using this to draft an email,” or “I’m using this to calm down for 10 minutes.” A purpose gives the session a boundary. Without one, the conversation can drift into emotional grazing.


If the purpose is emotional support, be honest about that. There is nothing wrong with wanting comfort. Just make sure AI is one support, not the only one.


Put time limits where your willpower fails


Willpower is weakest when people are tired, lonely, or upset. Use external limits.


Set a timer. Keep AI apps off your home screen. Avoid starting emotionally intense chats in bed. If late nights are the problem, create a cut-off time, such as no personal AI conversations after 10.30 pm.


A useful rule is to stop while the chat is still helpful, not when you are exhausted.


Pair AI use with one real-world action


For emotional topics, make the AI conversation lead somewhere outside the screen.


After discussing loneliness, send one message to a friend. After processing anxiety, take a short walk. After drafting a hard conversation, schedule it. After identifying a mood pattern, write it in a journal or bring it to a counsellor.


This prevents AI from becoming a closed loop.


Keep some questions for humans


Not every question should go to AI first. Some questions belong in relationships because the act of asking builds connection.


Ask a friend for their view. Tell your partner you felt hurt. Bring confusion to a mentor, teacher, elder, or therapist. Human support may be slower, but it carries something AI cannot provide: mutual life, shared history, and real accountability.


Notice the feeling after the chat ends


The aftertaste matters.


If you usually feel clearer, calmer, and more ready to act, your use may be supportive. If you feel foggy, dependent, lonely, guilty, or pulled to restart immediately, the habit may need adjustment.


Track this for a week. No judgement, just notes. Patterns become visible when they are written down.


Get help when AI becomes the only safe place


If AI is the only place where you can talk honestly, that is a signal worth respecting. It may mean you need more support, not less.


Consider speaking with a mental health professional, joining a support group, or reaching out to one trusted person with a simple message: “I’ve been isolating a bit and could use someone to talk to.”


You do not need to explain everything at once. One honest opening can start rebuilding connection.


A balanced future is still possible


AI conversations are compelling because they meet real needs. They offer speed, patience, privacy, and companionship in moments when human support feels far away. That is why people get attached. That is why the habit can grow quietly.


The answer is not panic or rejection. AI can be a useful tool, a creative partner, and even a temporary source of comfort. The healthier aim is to keep it connected to life rather than used as an escape from life.


Use AI to understand yourself, then return to the world with that understanding. Let it help you draft the message, but send the message to a real person. Let it calm you down, but do not let it become the only place you feel calm.


The best sign of balance is simple: after talking to AI, you feel more able to live your real life, not less willing to enter it.


We offer counselling, psychotherapy and life coaching to promote wellbeing and personal growth. Contact us to schedule your first session. Online session is available for anyone anywhere and we are located in CBD Singapore. Learn more or reach out for professional support at https://www.tlhcounselling.com/


 
 
 

Comments


bottom of page