Generative artificial intelligence, this friend who wants to devitalize our language

We now know that a long-practiced profession leads to “cerebral deformation”. Thus, the taxi driver will overdevelop his hippocampus, which stores the mental map of the cities he travels through; the musician, the auditory cortices; the perfumer, the olfactory cortices. Be careful, however: as in the last century, work lying down in a sawmill caused a pectus excavatumlethal collapse of the lungs, generative artificial intelligence (AI) is in the process of atrophying our faculty of language, the vital spring of thought.

Indeed, a living language makes it possible to say and interpret the world by freely adjusting one to the other. What’s more, it allows us to orient the framework by renewing forms, formulations and concepts. This is how language and society co-evolve. However, generative AI is fueled by what has already been said. It produces preformatted data following a pure probabilistic calculation. It systematically favors the forms most frequently encountered in the sea of ​​texts with which it is fed. She then re-serves us, in deceptively new packaging, only the expired product. Consequently, it devitalizes language and, insidiously, collectively ossifies our brains.

Meaning banished

As a career language expert, I was able to witness this dispossession of language firsthand. In this XXIe As the century draws to a close, we are witnessing an unprecedented change in the profession of translator. The imposition of AI-powered tools for the sole sake of efficiency leads the professional language specialist, postmodern Chaplin, to tirelessly tinker with texts made from “artificial” bits of language.

I had feared this hyper-technicalization of the profession for a long time. In the 1990s, during my doctoral studies, universities already swore only by “generative linguistics”. Get rid of the meaning! We then only had it for the form, the syntax. And just as in psychology, behaviorist ideology deemed subjectivity unworthy of science, an unobservable and unimportant interiority (horror!), linguistics shunned meaning and devoted itself to the quest for a universal grammar, a strictly formal code that was believed to be lurking and “programmed” in the brain, like a computer language…

Generative linguistics has since run out of steam, but the scientistic ideology that inspired it persists. The computerization of language has become the only avenue of “serious” research. Despite some pockets of humanist resistance, outdated, the translation was therefore destined for the AI ​​shredder.

The senses before the senses

In the 1980s, I was invited by my teachers to uninhibitedly read everything, absolutely everything, from the tube of toothpaste to encyclopedias, including literary and scientific works, the left and right press. I was thus invited to jovially place myself at the crossroads of human customs and knowledge. It is only rich in these “internalized” resources that I could aspire, one day, to translate “intelligently”, in tune with society.

Even more memorable, at the start of my career, my mentor, seeing me struggling on a technical text, told me: “To translate a text into electricity, you must see and feel the world like an electron. » This unusual advice contained a profound truth. At its core, to be understood, a language must be embodied, that is, lived in a body, experienced in the world. Likewise, to communicate, you must resonate with the subject and recipient of the message.

For the developing child, it is this organic and relational link that will give “meaning and body” to his or her mother tongue. However, what does a machine understand or feel when it generates the word “blue”? Nothing, nada ! Whereas for humans, this “blue” will first have been felt in front of a clear sky, a cornflower, a mother’s eyes… The meaning of blue, therefore, is first of all the blue experienced, only then blue says or writes.

For neuro-wisdom

In an era where AI invades all spheres of society, we urgently need to cultivate “neuro-wisdom”. I borrow this term from Idriss Aberkane, for whom the development of intelligence begins with healthy “neuro-ergonomic” practices. Let us first stop blindly entrusting AI, a black box without soul or conscience, with the responsibility of writing, speaking, and soon thinking for us.

There are, here and there, little glimmers of hope. A school recently returned to pencils to re-stimulate fine motor skills in primary school children, an ability which had been in decline since the widespread introduction of tablets in the classroom. It is this type of healthy withdrawal that must be encouraged and explored.

The cabinetmaker often puts down the tool and caresses the wood for a long time to tame it. The quality of his work depends on it. Let’s learn how to put down the computer to tame the language, without artifice. The quality of our thinking depends on it.

To watch on video


source site-47