🧱The superiority of the Internet system over the algorithmic system, because the first is open and the second is not

The debate held in Davos on 23 January 2026 bears on the articulation between the Internet system and the algorithmic system, known as AI.

One finds there the usual anthropomorphisms, against which one must nonetheless guard, whereby the issue would be to “educate” the software… and there is agreement on the contribution of these algorithmic tools, some maintaining that, because they “learn”, “speak” and “converse”, they might well constitute a space capable of replacing the space that the Internet constitutes.

But, setting aside the fact that the space of the Internet is occupied by persons and that software programs are not persons, the technical infrastructure of the Internet, through which the digital space was built, is open — which allows every person to gain access to it and to express themselves there.

That is not true of the algorithmic space, which is a set of closed places: open at first if the owner so wishes, free at first if the owner so wishes, then closed to whomever he wishes if he so wishes, or subject to payment if he so wishes — this being the question of law raised by the proceedings between the investor Elon Musk and OpenAI — whereas the Internet is structurally open.

Its very principle is access.

If the digital space were to be “rewritten” by the algorithmic system, presenting software as though it were human beings (agentic systems), then human beings — that is to say, real humans — would no longer have free access to the space of free communication.

Algorithms, which one does not educate, are only tools, and their status must remain that.