Biographical reference: Wikipedia
The question posed
Mr Illich, you distinguished convivial tools — those that enhance individual autonomy — from industrial tools that transform users into servants of the machine. Generative AI, accessible to all yet controlled by a handful of corporations, which category does it belong to?
The question is well put, but it contains a trap. A convivial tool, by my definition, is one you can use according to your own ends, at your own pace, in your own terms — without being taken in charge by an institution. Conviviality is not accessibility. A tool is not convivial simply because it is available to everyone.
The AI you describe is what I call a radical monopoly. Not because it is forbidden to anyone — on the contrary, it is offered to you freely. But because it progressively renders the alternative impossible: thinking for oneself, searching for oneself, constructing one's own answer by groping. When school convinces you that you need it in order to learn, you have lost something fundamental. When AI convinces you that you need it in order to think, it is the same loss — but at the level of thought itself.
What might have appealed to me, however, is the potential of the "learning webs" I had imagined — those decentralised networks where those who know meet those who wish to learn, without institutional intermediaries. If AI served to connect human beings sharing concrete knowledge, vernacular skills and living practices — it would then be convivial. But if it substitutes its own answer for that encounter, it is yet another counter-productivity.
The question is not technical. It is political: who controls the tool, who defines its ends, and who decides what "learning well" means?