The generative AI does not clarify communication: it clogs it. The noise theorized by Shannon becomes structural, making each message more difficult to perceive, to distinguish, to understand.
In the saturated universe of information, the rise of generative AI does not clarify messages: it multiplies its parasites. In the Chatgpt, Midjourney and others, the communication noise theorized by Claude Shannon reaches a new peak.
Originally, noise is a technical concept. Shannon, founding father of information theory, defines it as any alteration of the signal between the transmitter and the receiver. In a world dominated by machines capable of producing text, images, endless sound, this noise is no longer content to distort a message: it proliferates, it competes, it overwhelms. And he redefines the very field of communication.
A noise that has become structural
At the analog era, noise came from occasional interference – a parasitized telephone line, an illegible photocopy, a sizzling microphone. In the digital age, he changed in nature. It is no longer an accident, but an environment. The generative AI accelerates this phenomenon by multiplying issuers and homogenizing signals: everyone can publish, comment, generate credible content … but redundant, blurred, or empty.
A paradox sets in. The more we have access to sophisticated means of expression, the more difficult listening becomes. Information volume exponentially grows. According to IBM, 90 % of the world data has been created in the past two years (1). And among them, how readable, useful, intelligible?
Content automation and difference of difference
With generative AI, it is no longer necessary to master a language, style or logic to produce content. A prompt well turned is enough. Result: thousands of items, posts, videos or images are alike. Language becomes a foam, a bubble. Far from clarifying, this profusion disturbs perception.
Certainly, everyone can now publish elegantly. But this apparent democratization masks a deep standardization. AI engines are trained on the same corpus, recycle the same ideas, adopt the same turns. What we gain in accessibility, we lose it in singularity.
The gradual disappearance of “useful noise”
There was a time when the noise, even accidental, could be fruitful. A typing fault revealed involuntary truth. A slip revealed an unconscious. A disturbance made it possible to think differently. Today, the noise generated is neither poetic nor fertile: it is generalized, predictable, industrial.
In this context, even strong signals can become inaudible. Originality becomes suspect. Complexity, counterproductive. The human brain, already saturated, filters in priority what seems familiar to him. And algorithms, fond of similarity, strengthen this bias.
Towards algorithmic communication, without interlocutor
Another danger is looming: circularity. The AI generates content, which feeds other AI, which in turn generate. Soon we could attend automated exchanges without a human recipient. Pure, self -referential, self -sufficient noise.
This phenomenon is reminiscent of Facebook’s experience in 2017, where two bots had developed their own language, incomprehensible for their designers (2). Disconnected from all human interaction, communication then becomes a fiction. Or worse, an illusion.
Rethink the transmitter and receiver
Faced with this mutation, the question is no longer only: what do we say? But: Who? For what ? And in what mental or social space does it register? It is no longer enough to emit. It is necessary to rebuild a shared meaning, reinvent rituals of attention, slow down traffic to strengthen reception.
Shannon’s “noise”, yesterday, a secondary parameter, today becomes a first condition. Whoever knows how to reduce noise – or better, play with art – becomes a strategist. The challenge is no longer to speak stronger, nor more often. But to create silence around the message.
In a world saturated with generated language, the future belongs to communicators capable of restoring weight to each word, to restore links between voices, and to awaken attention in the din. Isn’t that the most human challenge?
Sources:
(1) IBM. How Much Data is Created Every Day?, IBM Cloud Blog, 2023. (2) Vincent, James. Facebook Shuts Down Ai System After Bots invent Their Own Language, The Verge, July 31, 2017. (3) Shannon, Claude. A Mathematical Theory of Communication. Bell System Technical Journal, vol. 27, 1948. (4) Lanier, Jaron. Ten Arguments for Deleting Your Social Media Accounts Right Now, Henry Holt, 2018. (5) Morozov, Evgeny. To Save Everything, Click Here: The Folly of Technological Solutionismem, Publicaffairs, 2013.
