At first glance, the case might seem trivial. It is anything but.
For Wikipedia is not some small literary journal championing the pen and purple ink. It is one of the most powerful symbols of digital knowledge. An encyclopaedia born of the internet, built collectively, constantly updated and consulted daily by millions of people.
In other words: the digital world itself is beginning to impose limits on the digital world.
The issue, incidentally, is not whether artificial intelligence can write. It can write. It can summarise, translate, rephrase, compare, synthesise and, in a matter of seconds, produce a text that would have taken several hours of work just a few years ago.
The real question lies elsewhere: who is accountable for what is written?
Artificial intelligence can produce a sentence that is perfectly credible and yet perfectly false. It can also reproduce an error, oversimplify a controversy or confidently present information from a dubious source.
Wikipedia is based precisely on the opposite principle: information must be open to discussion, sourced, corrected and attributed to human editorial responsibility.
This decision comes at a time when the debate surrounding AI is becoming much more serious. In recent days, Dario Amodei, CEO of Anthropic, has himself called for a slowdown in certain developments in artificial intelligence and for tighter controls over them.
We are therefore probably moving away from the first era of AI.
The era of wonder.
For three years, we have been asking: ‘What can it do?’
We are now entering a far more interesting era where the question becomes: ‘What do we still want to do ourselves?’
Writing is one such activity.
Not because a machine would be incapable of stringing words together correctly. But because writing is not merely about constructing sentences.
Writing is about making choices.
Choosing what to say, what to leave unsaid, what to verify, what we consider important. It sometimes means hesitating for twenty minutes over a sentence because a word isn’t quite the one we wanted.
A machine can mimic this hesitation.
It does not experience it.
Perhaps this is the magnificent paradox of this technological revolution: the more capable machines become of imitating us, the more they force us to define what, within ourselves, is worth preserving.
It is not, therefore, necessarily a question of declaring war on artificial intelligence. That would be as absurd as having declared war on the internet, the camera or word processing.
It is a question of deciding where we draw the line.
AI can be an extraordinary tool for researching, comparing, translating, unblocking an idea or exploring a lead. But if we gradually hand over the entirety of intellectual production to it, we risk one day finding ourselves in a strange world: machines writing texts for other machines, which will summarise them for humans who will never have read them.
Wikipedia has simply reminded us of something essential.
In the vast library of the future, we may need to continue reserving a few tables for humans.
