On March 12, 2026, LinkedIn published on its engineering blog the most detailed text ever written on how its algorithm works…
A technical program manager from the platform announces a change in doctrine. Distribution will no longer reward the size of the network, but the relevance of the message. Among the contents explicitly targeted by this recalibration is a newly named category, “generic thought leadership“, these featureless texts resulting from a crude copy and paste of AI tools; in other words: thought without anyone in it.
A confession disguised as a technical update
Most commentators saw it as a technical sanction against theartificial intelligence. This reading remains lazy in the face of a decision that is not. LinkedIn has actually just brought to light a problem that existed long before Generative AI do not take hold of it. This problem has a simple name: the absence of proper thought among those who speak.
And the figures confirm this diagnosis. According to the Algorithm Insights report, organic reach for active creators dropped 60% in two years. This shift confirms a fundamental movement initiated by the platform, which now favors the relevance of the statement on the size of the network, a shift already initiated by Instagram and TikTok a few years earlier.
This recalibration therefore does not target a tool, it targets a specific use of this tool. LinkedIn is not asking anyone to give up artificial intelligence to write their publications; the platform simply asks to stop letting her write for someone. This nuance seems minor on the surface, but it has long pleased those who preferred to ignore it.
The mistake we didn’t want to see
For years, the dominant belief in business communication has remained basic: exist on LinkedIn meant publishing regularly, regardless of the actual density of the subject, as long as the frequency of publication is maintained. The massive arrival of AI tools did not create this shortcoming, it only industrialized it and made it possible to produce, at almost no cost, texts that resembled expertise without ever incurring the intellectual responsibility of the person who signed them.
The most common analytical error is to place the problem on the technology side. The real problem actually lies with the order that everyone has placed on this technology. An AI that is asked to produce a point of view will produce an average point of view, a sort of statistical synthesis of everything that has already been said on the subject. This result does not constitute a defect of the tool, it corresponds exactly to its function. The emptiness that these contents are criticized for today corresponds to the emptiness that everyone explicitly ordered from them.
The real culprit is not the tool, it’s the silence we made it fill
The paradox deserves to be posed bluntly. Artificial intelligence has not impoverished managerial thinking, it has simply made visible a pre-existing poverty, at a speed and scale that no human laziness could have achieved alone. AI has not emptied the words of leaders of their substance; it only revealed its emptiness, at an industrial pace.
This observation shifts responsibility to where it truly should remain, with the speaker, and not with the tool that helps them formulate their ideas. A leader who has a real conviction, anchored in a decision taken or an arbitration assumed, will produce an identifiable text even when assisted by artificial intelligence. A leader who has nothing to say will produce emptiness, whether he uses this tool or not. LinkedIn has simply made this void impossible to monetize in visibility.
What becomes non-negotiable
This shift has a direct consequence for anyone who speaks publicly: rarity once again becomes a decisive advantage. In a saturated flow of interchangeable content, what distinguishes a leader is no longer the frequency of his publications, but his ability to formulate a precise disagreement, to quote an exact figure rather than a vague generality, and to assume a position which will not necessarily achieve consensus. This requirement requires time and a form of personal exposure, as well as the acceptance that a good text truly engages its author much more than it protects him.
It is therefore not artificial intelligence that we must question today, but the demands that everyone places on using it. The question has never been about writing alone or with assistance. It has always focused on the existence, at the time of publication, of truly personal content to transmit. LinkedIn has just made this question impossible to dodge any longer, and this development is good news for anyone who still has something to say.
