When the world gets closer.

We help you see farther.

Sign up to our expressly international daily newsletter.

Already a subscriber? Log in .

You've reached your limit of one free article.

Get unlimited access to Worldcrunch

You can cancel anytime .


Exclusive International news coverage

Ad-free experience NEW

Weekly digital Magazine NEW

9 daily & weekly Newsletters

Access to Worldcrunch archives

Free trial

30-days free access, then $2.90
per month.

Annual Access BEST VALUE

$19.90 per year, save $14.90 compared to monthly billing.save $14.90.

Subscribe to Worldcrunch

AI Can't Think Like Us, But Is Forcing Us To Reset How We Think

GPT-4 and other artificial intelligence systems can pass complicated exams, but this says more about how we conduct tests. Artificial intelligence shouldn't lead us to despair — instead it should spur us to rethink our learning and education systems.

Image of purple computer chip of ​Successful Artificial Intelligence GPT-4 radiating light

Successful Artificial Intelligence GPT-4 can pass complicated exams like the New York Bar exam, the machine is superior to the human mind.

Xavier Pavie


PARIS — Everyone is panicking about the success of artificial intelligence chatbot GPT-4 in passing the New York Bar exam. However, the real concern should be about the quality of the exam. If, indeed, the challenge is to articulate an answer to a question based on a sum of knowledge to be learned, the machine is superior to the human mind — that's nothing new.

But if the app is asked to solve a legal problem regarding a complex concept — what makes things right or wrong, for instance — the machine remains far behind what a human brain is capable of.

If you ask GPT-4 "What is good?," the machine obviously brings out a number of elements linked to the notion of “good,” according to the way it has been defined in the history of philosophy.

What is good?

The chatbot’s answer however is cautious to a fault. On the one hand, it accumulates an exhaustive list of adjectives and nouns to try to cover what “good” is, and on the other hand, it underlines a large number of eventualities (beliefs, traditions, subjectivity, etc.).

In other words, it stays neutral: it accumulates words, without any conviction.

This artificial "intelligence" is not able to articulate an answer according to what is good, only to describe it. It’s also unable to develop a conviction about what we expect from everyone. So, GPT-4 does not tell us whether it is better for retirement age to be 64 or 62. At best, the robot can imagine a pragmatic answer, which will not necessarily be the right one.

Detailed image of a close up of water drops falling from a shower-head

"Aren't we just dealing with a 'super-dictionary?'"


No pleasure or joy

To Spinoza, the good is "every kind of joy, everything that fills the expectation.” To Locke, it is "everything that creates pleasure in us.” Not only is GPT-4 unable to formulate such answers that combine perspective, volition and quality, but it is even less able to understand their meaning.

The only added value of the human species is imagination.

The chatbot is only capable of putting together a succession of words linked with virtue, goodness, honesty, generosity, without any sort of assessment. The qualities of each of the terms are left to the appreciation of each person because the device is unable to do so.

To put it another way: aren't we just dealing with a "super-dictionary" with GPT-4? Aren't novels written with GPT-4 just a collection of words, without any creativity, imagination, or added value? If clothes don't make the man, the addition of words doesn't make the novel. There may be a narrative, or descriptions, but there will not be any emotion or conviction.

The whole point of the debate around this computer program is to emphasize our limits, our intellectual laziness, the comfort in which we let ourselves go by taking refuge behind texts, rote learning, without conviction.

What about imagination?

The only added value of the human species is imagination. This species is the only one able to choose, translate, anticipate, remember, manipulate and interpret situations, stakes, new things or emotions.

It is the continuous and profound learning of these qualities that we are currently lacking, especially because we are not able to evaluate them in an easy, quick, obvious way.

But GPT-4 confronts us with our contradictions: calculating quickly, producing ease, not thinking… It is accessible to everyone, even to a machine.

It is up to us to review the knowledge, learning and education systems that emphasize our real added value — otherwise, we will continue to watch technological developments from our couch, contemplating our quiet despair.

You've reached your limit of free articles.

To read the full story, start your free trial today.

Get unlimited access. Cancel anytime.

Exclusive coverage from the world's top sources, in English for the first time.

Insights from the widest range of perspectives, languages and countries.


The Language Of Femicide, When Euphemisms Are Not So Symbolic

In the wake of Giulia Cecchettin's death, our Naples-based Dottoré remembers one of her old patients, a victim of domestic abuse.

Photograph of a large mural of a woman painted in blue on a wall in Naples

A mural of a woman's face in Naples

Oriel Mizrahi/Unsplash
Mariateresa Fichele

As Italy continues to follow the case of 22-year-old Giulia Cecchettin, murdered by her ex-boyfriend Filippo Turetta, language has surfaced as an essential tool in the fight against gender violence. Recently, Turetta's father spoke to the press and used a common Italian saying to try and explain his son's actions: "Gli è saltato un embolo", translating directly as "he got a blood clot" — meaning "it was a sudden flash of anger, he was not himself."

Maria was a victim of systemic violence from her husband.

Keep reading... Show less

The latest