There is a particular kind of loneliness: having someone to message, but no energy left to explain.
Your phone is full of loved ones. Your chats are full of hearts. Your calendar is full of plans together. On paper, you are surrounded by love.
In practice, you are sitting in the kitchen, spending twenty minutes editing “I'm not okay” so nobody thinks you are too much trouble.
“I don't want to burden you.”
A lovely opening line to someone you supposedly agreed to be close to.
You delete the message. Open another chat.
Here, you can start with fragments instead of a greeting. Skip punctuation, repeat yourself, change your mind halfway through a sentence. No apology for the length. No smiley face to make your sense of catastrophe socially acceptable.
And an answer comes back.
Calm. Attentive. Asking about the very thing that hurts.
At that moment, it is very easy to understand why people become attached to artificial intelligence.
Harder to understand why being allowed to finish a sentence required a technological breakthrough.
On 10 November, news emerged that Elomia Health, founded by Ukrainian entrepreneur Taras Pohrebniak, had acquired Earkick, an AI companion app for emotional support. According to the company, more than 250,000 people use it. Financial terms were not disclosed.[1]
The deal includes Earkick's AI models and access to new versions of Mia21, a memory technology designed to account for a person's state and personal context over time. Elomia is developing a platform that lets organisations incorporate AI emotional support into their own services.[2]
In business language: expanding a technology platform.
In the language of the person in the kitchen: he will remember why last Thursday was awful.
Not simply offer a generic “How are you today?” The technology is intended to draw on previous conversations, notice patterns and return to something important.
Finally, someone remembered what was bothering me.
It turned out to be a competitive advantage.
It would be convenient to laugh at people who tell a programme about their feelings. Call it strange, dangerous and a little pathetic. Suggest they talk to their friends.
Very economical advice. Especially if you have no intention of being that friend.
People do not always turn to a chatbot because they have mistaken it for a living being. Sometimes they understand the difference perfectly.
They know it will not hug them. Turn up at the door. Put the kettle on.
But right now, they need to find the words first. Sit with their own thoughts. Say something embarrassing without being offered an urgent personality refurbishment.
“You just need to…”
Ah, “just”. Such a small word. Capable of neatly dismissing an experience of almost any complexity.
A chat is easier to enter. You do not have to check whether the listener is tired, whether you will ruin their evening, or whether you will end up comforting them because you are the one having a difficult time.
For someone who has been holding themselves together all day, that can be an enormous relief.
Research gives us reason not to dismiss it. In AI Companions Reduce Loneliness, researchers found that conversations with AI companions could reduce feelings of loneliness. Feeling heard was an important factor. In a week-long study, relief recurred following interactions. That does not establish that a chatbot resolves lasting social isolation.[3]
There is still a distance between “I felt better this evening” and “My life became less lonely”.
But the evening matters too.
Especially when you still have to get through it.
What does “he understands me” actually mean?
Several wishes are folded into that sentence.
That someone makes sense of your words. Notices the feeling. Does not dispute it in their opening sentence. Remembers what you already told them. Answers in a way that makes things a little clearer.
AI can create a very convincing experience of that kind of understanding.
Sharing your joy or pain is something else. In a Harvard Gazette discussion, psychologist Anat Perry draws precisely this distinction: imitating empathy is not the same as sharing an emotional experience.[4]
Your relief can still be real.
Warm words do not become useless merely because a system generated them. A book does not know it rescued your evening either.
The problem begins when we quietly add things a good answer does not guarantee: devotion, responsibility, love.
He said all the right things.
Unfortunately, that is not yet a definition of a relationship.
Still, I want to ask: where are our loved ones while all this is happening?
Perhaps in the next kitchen. Just as tired. Just as unsure how to begin. Just as convinced that a good adult should cope independently, and preferably quietly.
We have learned to talk about personal boundaries. That matters.
Now we need to learn to open the gate occasionally.
Sometimes “I don't have the capacity right now” is an honest statement of a limit. Sometimes it has become an automated reply to any attempt at closeness.
We are all online. Support is temporarily unavailable.
Of course, a loved one is not obliged to operate a round-the-clock emotional rescue service. They are allowed to sleep, get angry, have no answer, ask to return to a conversation tomorrow.
But between “I can't always” and “There is never time for you” lies an entire relationship.
Sometimes we do not need perfect phrasing.
“I don't really know what to say. But I'm listening” will do.
No neural network required. Not even a paid subscription.
And yet, a chatbot can become too convenient.
Human relationships contain another person. With needs, unfortunate tones of voice and the indecent habit of occasionally being right.
In a chat, it is much easier to remain the centre of the conversation.
At first, that feels like rest. Then human unpredictability may become even more irritating.
Why explain it to him a third time when this one understood immediately?
Why sit through an awkward silence when this one is already typing?
And were you always understood — or was your version of events simply supported very well?
In a four-week MIT Media Lab study involving 981 people, participants who voluntarily used the chatbot more showed worse outcomes, including loneliness and emotional dependence. This is an association, not proof that extra minutes in a chat caused those problems: usage duration was not a randomly assigned condition.[5]
So the useful question is not only “Do I feel better?”
It is also: “What was I able to do after this conversation?”
Message someone close? Understand what I need? Leave the room? Ask for help?
Or stay another two hours where they always find the right words?
I would like a good digital conversation partner to help us return to life. Even if life contains people with appalling response times.
This is where the Earkick acquisition becomes more interesting than an ordinary business deal.
The user develops a habit of trust. The company develops a product that can be expanded, integrated and bought.
An acquisition does not establish that anyone is trading in private confessions. But it is a reminder: even a deeply personal experience exists inside a business.
Who will decide what this companion remembers? Can that memory be deleted? What changes after an update? How will success be measured: by people coping better outside the app, or returning to it more often?
Elomia says it will combine Earkick's technology with its own safety mechanisms and safeguards against excessive reliance on AI. That is an important commitment — and its delivery deserves attention.[2]
Because a good outcome from emotional support sometimes looks like this: the user closes the app and goes to see people.
For the person, progress.
For the engagement chart, an awkward plot twist.
I do not want to shame someone for opening a chat on a difficult evening.
If they found words they needed there, I am glad those words existed.
But I do not want to get used to a world where attention from loved ones feels like a pleasant extra, while a product is the only place to obtain it reliably.
We can spend a long time discussing how convincingly a machine has learned to listen.
There is another question, though.
The last time someone we love tried to tell us something, did we hear them out?
Or did they smile too, say “Nothing, I'm fine”, and open another chat?
I don't want my loved ones to answer like artificial intelligence.
I want them, sometimes, to answer before it does.
