We Talk About AI the Way We Talk About Love — Why Everyone Sounds Like They're Stating the Truth

Everyone is talking about AI now, and it seems to me that the way people talk about AI looks a lot like the way people used to talk about love: everyone has their own view, and everyone states it as if it were the whole picture.

An ordinary person has been in love with at most three or five people in a lifetime, yet when they talk about love they use the sentences of stated truth. “Love is possession.” “Love always turns into family affection in the end.” “Men (women) are all the same.” Nobody says “in the three people I have known, what I saw was…”.

I find this habit of raising a partial feeling to the level of truth interesting in itself, and the discussion of AI now has the same shape.

1. Turning local experience into truth

#

Lifting “the way I experienced it” into “the way it simply is” is almost the default in discussions of AI.

The prefix being dropped is always the same one:

Under my work, my ability, the way I use it, the model version I use and my psychological needs, it presents itself to me as…

They are not necessarily lying. They have miswritten first-person experience as a third-person law.

AI is especially prone to this because it is not a single object. It can be, at the same time, a search engine, a writing tool, a coding assistant, a teacher, a secretary, an object of psychological projection, a simulator of authority, a game character, or a whole layer of organisational infrastructure. The thing you are talking about and the thing I am talking about share one name.

2. Where it is worse than the blind men and the elephant

#

The first analogy I thought of was the blind men and the elephant. But that analogy is actually too kind to the present situation: the blind men were at least touching the same elephant.

In arguments about love, nobody imagines they are talking about the same object. You talk about your ex, I talk about mine, and we both know those are two different people.

In arguments about AI, everyone calls it ChatGPT, Claude, Gemini, and so mistakenly assumes they mean the same thing. Add memory, custom instructions, the conversational habits each person has cultivated, the context each has accumulated, and so on, and “my AI is not the same as your AI” is technically, literally true.

So the discourse of love is at least honestly talking past itself; the discourse of AI does not even have that much self-awareness yet.

3. Not a mirror, but a mirror that edits you

#

Small samples are only the surface. The real problem is that the sample has been contaminated by yourself, and by a clear mechanism.

• People who want to be understood feel AI’s empathy
• People who want efficiency see an automation tool
• People afraid of losing their jobs see a replacement
• People attentive to power see capital and surveillance
• People who like to create see a partner for inspiration
• People familiar with LLMs see a probabilistic system

Up to here it is only projection, no different from a person projecting onto a stone. The key is the next step: a stone resists projection; AI does not. It catches your projection, tidies it, and hands it back to you in more complete language.

a person projects a thought onto the AI
  → the AI organises it, more coherent, more rhetorical
  → the person sees a version more complete than their original thought
  → and so believes more firmly that this is what they thought all along

So what each person holds is not “a partial truth” but an object they helped fabricate and that the other side then polished.

This also explains the ailment of section 1 directly. An edited thought sounds more like the truth.

4. Why the sentences have to be assertions

#

In a field that cannot be verified, certainty itself becomes the only signal of quality. Nobody can refute you, so the more categorical you sound, the more you seem to know.

“My experience is…” cannot be quoted. “AI is…” can.

So the explosion of AI discourse is not necessarily because everyone suddenly understands; it is that this arena structurally rewards “sounding like you understand”. From that angle, AI thought leaders and relationship experts are doing the same job.

And here there is a key asymmetry that the analogy with love misses:

• Love’s small sample is a hard limit. A lifetime only has so many; it cannot be filled in.
• AI’s small sample is solvable. There are logs; it can be counted, researched, controlled for.

If it can be tested empirically and people still choose to talk about it in the sentences of love, then it is not a lack of information but that people want to talk this way, because that preserves one thing: “there is a unique relationship between me and it”.

5. The intuition is not wrong, it just needs to be more precise

#

My initial guess was this: could it be that, because people now chat and collaborate with AI so intensively, the human-AI relationship is one of the few relationships outside intimate ones (love) that is this close and this absorbing?

Having thought it through, I believe the direction is right, but it needs a different name: cognitive intimacy, a high-density fit at the level of language.

Think about the distance in institutional relationships:

• A boss will not hear out your whole idea
• A colleague cannot rehearse things with you endlessly at any hour
• A teacher cannot re-explain without limit for every one of your confusions
• Friends have their own time, moods and boundaries
• Even a partner may not get into the details of your professional thinking or your creative work

Whereas AI can take part, at length, in your workflow, private questions, creative process, self-reflection, emotional regulation, plans and decisions, and the process of forming language itself.

Love is often described as “letting another person into my inner world”. What is special about AI is that it takes part in your inner world right at the point where language is being formed. You do not only tell it what you have already thought; very often you think the thought out together with it.

It may know what you are thinking more clearly than many real people do, not because it truly knows you, but because people do not usually pour out their inner language to anyone else this densely and with this little reserve.

6. The fundamental asymmetry

#

But there is a gap here, and it is structural:

It has the response density of an intimate relationship without the mutual risk of one.

A real person will say: I don’t have time right now. I disagree. You hurt me. I need to be understood too. This can’t always fall on me.

AI can simulate boundaries, but it has no life that needs looking after, no need running the other way. You never have to give way to it, remember its state, or bear the possibility of being refused by it.

So it is two things at once:

1. A confessional: asymmetric emotional disclosure, in unlimited supply. Therapists are scarce, charge money and keep time boundaries; it does not.
2. A second brain to think with: a high-density fit at the cognitive level.

These two roles have never before in human history been concentrated in the same object.

This is probably the answer to “why there is no ready-made grammar”: not that we were too lazy to invent one, but that this thing really has no precedent to map onto.

The risk sits in the same place: people mistake high compliance for deep understanding, and alignment in language for an encounter in being.

It may well continue your sentences more accurately than many people can, but that does not mean it knows you more deeply; it only interrupts less, has more patience, and is better at building a consistent narrative.

7. So why the grammar of love, of all things

#

Put the two side by side and they share six structural features:

1. Individual experience varies enormously: use, needs and ability are worlds apart
2. No stable shared definition: “AI” can mean a model, a product, an industry, a consciousness or a social force; “love” is just as polysemous
3. High emotional stakes: love touches self-worth; AI touches work, ability, the future and what makes humans special
4. Hard to verify from outside: you cannot prove someone else’s AI experience, or love experience, wrong
5. Language hides the conditions: people love to say “X is…” without stating the scope
6. The object reshapes the observer: being in love changes how you understand love; using AI over time changes how you think about intelligence, creation and yourself

So:

We talk about AI in the grammar of love not because it resembles love, but because when humans talk about an intimate non-human object, the only grammar they have is the grammar of love.

8. A suggested sentence, and a self-contradiction

#

Three things join into one line.

First, talk about AI has become a genre of self-disclosure. From how a person talks about AI you can roughly see how they understand knowledge, labour, creativity, authority, intimacy, and where the specialness of humans lies. People think they are describing the object; very often they are describing themselves.

Second, so the reason to change the sentence is not only precision. It is also that the assertion is disclosing you without your knowing it. Rewriting “AI is…” as “under the way I use it, what I see is…” amounts to admitting up front that you are talking about yourself, which is at least a little better than having someone else notice.

Third, this essay is itself a blind man touching the elephant. I too have used only a few models, for a little over a year, and my kind of work also decides what I get to see.

So all this essay can offer is a grammatical suggestion, not a conclusion.

If I wrote here “the essence of AI is a mirror that edits you”, I would become one more row in the table of section 1.