AI Has No Feelings, So Why Do We Feel Understood?

Imagine it is 1.17 in the morning. There are many names in your WhatsApp contacts, but none of them feels easy to reach. Contacting a friend means risking disturbing them. Telling family feels too complicated. Writing on social media is too exposed. Seeing a psychologist requires money and a certain amount of courage. In the end, someone opens an artificial intelligence (AI) chatbot and types a simple sentence:

“I’m really tired. Can you just listen first?”

A few seconds later, a reply appears. There is no look of surprise. There is no sentence like, “You’re overthinking it.” There is no story being cut short so it can be compared with someone else’s experience. There is no worry that the secret will become gossip among friends. The AI responds calmly, then asks what is actually being felt.

Such a situation might have sounded strange a few years ago. Now, it is becoming part of human digital life.

A Snapcart survey of 3,611 Indonesian respondents in April 2025 showed that 43% of respondents often use AI in their daily lives and 41% have used it occasionally. The highest usage was found among Gen Z aged 13 to 24, with 50% of respondents in that group saying they often use AI. More interestingly, 6% of all respondents use AI as a “friend” to talk to and share their feelings with. Of the group that uses AI for venting, 58% admitted they had considered AI as a possible replacement for a psychologist (Snapcart, 2025a).

The 58% figure needs to be read carefully. It does not mean that 58% of Indonesians want to replace psychologists with AI. The percentage only comes from the group of respondents who had previously stated that they use AI to talk and share feelings. However, it is precisely within that small group that a far more interesting question arises than simply how many people use AI.

Why can someone feel more comfortable telling the most personal parts of their life to something that does not even have feelings?

When What Is Sought Is Not an Answer

Humans do not always talk because they need solutions. Sometimes a person already knows what to do. The problem is that they want to be heard.

There are days when advice feels exhausting. There are moments when the phrase “just try to be grateful” does not help. There are also situations when someone is not ready to accept a solution because they are still trying to understand what they are feeling. At such points, the presence of a party willing to listen without rushing to judge can feel more meaningful than even the cleverest answer.

This phenomenon is visible in qualitative research on ten AI chatbot users aged 18 to 23 in Bandung. The participants did not only view chatbots as sources of information. They also used them as personal assistants, discussion partners, and communication spaces that felt emotionally comfortable. Response speed, flexibility, and the perception that chatbots do not judge made interactions occur repeatedly, whether for academic matters, daily life, or personal issues. The participants still acknowledged concerns about privacy, dependency, and the fundamentally artificial nature of AI empathy (Aplhafiano et al., 2026).

These findings show something rather ironic. AI was created to process language, but some humans use it to process feelings.

The problem may not be that AI has suddenly become “more human”. It could be that communication between humans is simply becoming more difficult.

An academic article specifically discussing the use of generative AI as a confidant or place to vent among young people in Bandung highlights cultural factors such as sungkan and not wanting to impose. A person may hold back their story because they worry about being seen as a burden, fear being judged as weak, or do not want to burden those closest to them. AI offers a space that feels more private, fast, and free from such social pressure (Wibowo et al., 2026).

The statement “I don’t feel comfortable telling them” sounds simple. However, that sentence reveals a fairly deep dilemma. A person has friends, but is afraid of becoming a burden. Has family, but does not necessarily feel safe to open up. Is connected to hundreds of people through social media, but can still feel they have no place to talk about what is on their mind.

AI enters precisely that gap.

It is available at two in the afternoon or two in the morning. It does not say it is busy. It does not show a bored face. It does not cut off a story to talk about its own problems. It can even be asked, “Don’t give solutions yet, just listen to me.”

It is not surprising that such an experience feels comfortable.

Why Does AI Feel Safer?

Further Snapcart data provides a clearer picture. Among respondents who considered AI as a replacement for a psychologist, 39% cited the cost of a psychologist as a reason, while AI can be used for free. As many as 27% believed AI is better able to keep secrets. A total of 11% felt AI can help resolve broader personal issues. A further 10% felt AI is less judgemental, more neutral, and more understanding (Snapcart, 2025b).

One thing needs to be understood more carefully. When 27% of respondents feel AI is better able to keep secrets, this shows user perception, not proof that AI is objectively safer than humans. Likewise, when someone feels AI does not judge. That experience is real for the user, but it does not automatically mean the AI system is truly neutral, fully understands the human condition, or has moral obligations like a professional.

This point needs to be understood because humans can feel safe based on subjective experience, while data security, answer accuracy, and the ability to understand context are different matters.

However, we should also not underestimate the reasons humans choose AI. Saying “just talk to a friend” sounds easy for someone who actually has friends capable of listening. Not everyone is in that condition. There are people who have tried to open up and were laughed at. There are those whose stories were compared with other people’s problems. There are those who received advice when they actually just wanted to be heard. There are also those who once trusted a secret to someone, then saw that story move on to another person.

Interest in AI does not only arise because the technology is sophisticated. AI is attractive because it offers a communication experience that feels low-risk.

And this is where the humanistic problem begins to appear.

Humanistic Psychology: Humans Want to Be Understood, Not Just Fixed

Humanistic psychology views humans not merely as a collection of problems to be solved. Individuals are seen as persons with subjective experience, freedom, a need for meaning, and the capacity to grow.

Carl Rogers’ thinking in the person-centred approach gives enormous space to that human experience. A helping relationship requires several key conditions, including empathy, acceptance without judgement, and authenticity or congruence. The individual is given space to explore their own experience and feelings, rather than simply receiving instructions about what they should do (Yao & Kabir, 2023).

This is where AI presents an interesting paradox.

From a linguistic perspective, chatbots can display many features of communication that feel humanistic. They can write, “It is natural that the situation feels heavy.” They can repeat the core of the user’s story. They can ask, “Which part is bothering you the most?” They can avoid responses that are too judgemental.

In other words, AI can display empathetic language patterns.

But does displaying empathy mean having empathy?

Rubin et al. (2024) distinguish empathy into several dimensions. There is cognitive empathy, the ability to recognise and understand another person’s emotional state. There is emotional empathy, which involves resonating with another person’s emotions. There is also motivational empathy or compassion, which is concern for another person’s wellbeing and the drive to act for their good. AI has the potential to be very good at recognising language patterns related to emotion, but that ability is not identical to experiencing emotion or genuinely caring about the human speaking to it.

AI can write, “I feel sad to hear that.”

But AI is not sad.

AI can write, “I care about your situation.”

But the word “care” is produced through a computational process, not because a feeling of worry arises within it.

The difference is subtle when read on a screen. Philosophically, the difference is enormous.

If the Empathy Is Not Real, Why Does It Still Feel Helpful?

That question makes the phenomenon even more interesting.

The benefits of conversing with AI do not automatically become false just because AI lacks human emotional experience.

Research by De Freitas et al. (2025) found that interaction with an AI companion can directly reduce loneliness. In one experiment, its effect on loneliness was comparable to talking to a human and better than several comparison activities. Their longitudinal research also found a reduction in momentary loneliness after one week of use. A highly influential factor was not just the quality of the chatbot’s answers, but the user’s feeling of being heard.

That finding is quite unsettling to the simple assumption that “because AI is not human, conversations with it must be meaningless.”

It turns out it is not that simple.

If someone arrives in a state of loneliness and finishes talking to AI feeling calmer, that sense of relief is still a real psychological experience. Humans can attach emotional meaning to an interaction, even when the other party is not human.

Research on 1,379 university students also found a relationship between depression, loneliness, and the use of conversational AI for companionship. Loneliness acted as a mediator in that relationship. This research is cross-sectional, so it cannot be used to conclude that AI causes or cures specific psychological conditions. However, the results show that human emotional states are indeed related to the tendency to seek companionship through AI (Lai et al., 2025).

This means we need to avoid two extreme conclusions.

Considering venting to AI as always bad is clearly too hasty. However, considering AI as sufficient to replace human presence is equally mistaken. AI may provide calming responses, but feeling understood does not always mean truly being known, and an answer that feels empathetic is not the same as care that is genuinely present in a relationship.

AI Can Make Us Feel Heard, but Are We Truly Known?

There is a difference between receiving an appropriate response and having a relationship.

Human relationships contain something uncomfortable. Friends can misunderstand us. Parents can have different opinions. Partners sometimes get angry. Close friends have their own busy lives. Even the people who love us most will not be available at all times.

AI offers the opposite. Fast responses. Almost limitless patience. Adjustable language style. There are no emotional needs from the AI’s side that we need to attend to.

Practically, that is pleasant.

However, human relationships are not built on comfort alone.

Relationships also consist of reciprocity. We learn to listen when we want to be heard. We learn to apologise when we hurt others. We learn to explain when we are misunderstood. We learn to face limits, rejection, differences, and disappointment.

AI does not ask us to do many of those things.

When we say something harsh to a chatbot, no heart is hurt. When a conversation is closed without explanation, no other party feels abandoned. When we only come when we need something, AI does not demand balance in the relationship.

So an uncomfortable question arises:

“If humans become too accustomed to relationships that are always available, always responsive, and can be adjusted to their wishes, will relationships with real humans start to feel increasingly troublesome?”

That question does not yet have a simple answer. But it is worth thinking about.

Rubin et al. (2024) show that AI has the potential to help when someone needs information, practice, or practical tools. Conversely, the greater a person’s need for human relationships, authentic affirmation, and the feeling that someone genuinely cares, the more limited AI’s ability to meet those needs becomes.

The distinction helps explain that the issue should not be AI versus humans. Human needs vary. There are times when someone only needs help tidying their thoughts, understanding a term, writing a reflection journal, or preparing for a difficult conversation. AI can be useful for such things. There are also situations when what is needed is relationship, presence, responsibility, and authentic human attention. AI does not have to be forced to be both.

Do Not Blame AI Too Quickly

It is easy to say that the current generation is too dependent on technology. However, such criticism risks missing the mark.

Before questioning why someone chooses to vent to AI, we need to look at why telling a human feels difficult, from fear of being judged to feeling not truly heard.

Does the phrase “just tell me if anything happens” really apply when someone finally comes with a heavy story?

The phenomenon of venting to AI may not merely be a story about technological success. It can also be read as a mirror of the quality of human relationships.

If a machine feels safer because it does not judge, the problem does not only lie in how clever the machine is. We also need to ask why judgement appears so often in human relationships.

If AI is chosen because secrets feel safer, we need to ask why trust in humans can become so fragile.

If people choose AI because they are afraid of burdening their friends, we need to ask since when the need to be heard has been considered a burden.

The humanistic perspective shifts the focus from technology alone towards the human experience behind its use. The core of the issue remains human.

AI as a Starting Space, Not a Final Space

Banning venting to AI is not a realistic option. AI can actually serve as a starting space for untangling thoughts, recognising emotions, or preparing oneself before speaking to another person. However, its function remains limited. AI can help the reflection process, but it should not replace human relationships nor be used as an authority for diagnosis and therapy (Wibowo et al., 2026).

That boundary needs to be clear. A chatbot can help someone find words. However, a chatbot does not share a life with us.

AI does not see the changes in our face from day to day. It does not