Why does ‘I read it in a book’ sound smarter than ‘I asked AI’?

There is a strange social asymmetry.

Someone says: “I once read about this in a book.” Which book, who wrote it, what data the conclusion rests on and whether the person remembers it correctly, often nobody asks: the mere fact of a book already sounds respectable. But say “I asked an AI, requested sources and checked the answer”, and in some listeners an entirely different reaction switches on: right, so you cannot think for yourself.

Even though, by the quality of the result, it may be exactly the other way round.

This is more than a personal impression

In 2025 Scientific Reports published a series of five preregistered experiments, and in one of them people were shown the same piece of advice written by ChatGPT. One group was told it had been created by ChatGPT, the other that it was another person’s answer, and when the origin was linked to ChatGPT, the identical text was rated lower for quality, effectiveness and sincerity.

Another 2025 study, published in PNAS, ran four experiments with more than 4,400 participants. People who used AI at work were on average judged lazier, less competent and less diligent, and the participants anticipated that attitude in advance and therefore had reason to hide their use of AI. It is worth pausing here, because this cost has an addressee: a worker who hides the tool deprives the employer of knowledge of how the work is actually done, and therefore of the ability to check it. The stigma does not remove AI from the process; it removes it from view.

So the jibe “ChatGPT wrote that for you” does not merely exist in comment sections; it already has a perfectly measurable social price.

Perhaps we are judging the cost of the answer, not the answer itself

The chain used to look roughly like this:

knows a lot → has read a lot → spent years → a competent person.

AI breaks that link. A first overview of a topic, a comparison of versions and a list of sources can now be had not in a few days but in a few minutes, and the observer gets the feeling that the result came too cheap.

This resembles what is called the effort heuristic. In classic experiments people rated poems, paintings and other works more highly when they believed more time and labour had gone into them, and the effect became especially strong when quality was hard to judge directly.

Something similar happens with information. When we cannot quickly check the argument itself, we start judging by indirect signs: the thickness of the book, years of education, a confident tone, the amount of time spent. AI removes the familiar sign of effort, so the sign of quality seems to disappear with it, but that is a poor substitute for checking: if person A spent four hours looking for the right answer and person B got the same answer in ten minutes with the help of AI, checked the primary sources and understood the reasoning, the informational value of the result does not shrink merely because the route was shorter.

A book is also a status symbol

The phrase “I read about it in a book” reports more than a source. It carries a whole cultural trail: education, discipline, seriousness, the habit of long text. The phrase “I asked ChatGPT” still often carries a different trail: did not know, did not look into it, pressed a button, copied the ready-made.

Both pictures can be false. One can have read outright nonsense twenty years ago and remember it even worse, one can open the first book to hand by an author who has simply packaged personal convictions attractively. Or one can use AI as a research tool: ask for several explanations, find contradictions, get a list of primary sources, open them and check what is actually written there. Of course the reverse is possible too, thoughtlessly copying the model’s first confident hallucination. The delivery method of information does not say whether intellectual work was done.

It is also worth asking whom this asymmetry suits, and here we leave research data for reasoning. The years spent reading were not only a way to learn but an entry ticket: into a profession, into a conversation, into the right to judge. Whoever paid for the ticket has no interest in the entrance getting cheaper, and the jibe “you cannot think for yourself” protects not the quality of thinking but the value of an investment already made. This is not an accusation; any group whose status rested on a rare skill would react the same way; it is simply useful to understand that the irritation is aimed at the price, not at the answer.

“AI said so” is still a weak argument

Defending a new tool, it is easy to fly to the opposite extreme. AI does not become an authority simply because it answers at length, quickly and confidently: a model can invent a fact, confuse a source, miss an important exception or adapt itself to the wording of the question. Alongside aversion to algorithms there is the opposite problem, excessive trust in automated advice even when contradictory information is available. Where such trust leads once it becomes the norm and disobeying the system becomes irrational is covered in the article on invisible steering.

So the phrase “it is true because ChatGPT said so” is bad. But the phrase “it is true because I read it in a book” is, by its logic, no better.

A book, a search engine, a teacher and a language model can all help along that path, but none of them frees a person from the need to understand what exactly they are repeating. Where normal delegation ends and loss of control over the result begins is the subject of “Delegation is not degradation”.

What is becoming valuable now

What is gradually becoming scarce is not access to information but the ability to pose a good question, spot a contradiction, choose a criterion, check a source and apply the conclusion to a real situation.

The skill of remembering hundreds of facts will not disappear, but on its own it no longer gives the old advantage, because AI has made search, first-pass overview and the compilation of information cheap. That is exactly why it causes so much irritation: it cheapens actions that long served not only as tools of work but as marks of intellectual status. Whether the market pays for what has become cheap, and what it pays for instead, is covered in the article on AI slop.

The question is no longer whether a person used AI. The question is what they did after the answer.

Understood it? Checked it? Compared it? Saw the weak points? Or simply pasted in a pretty text?

That is what deserves judgment.