The large language model is the most powerful knowledge-processing system ever constructed. It retrieves, organizes, synthesizes, and generates formal knowledge with a speed and comprehensiveness no human can match. But it does not possess intelligence in Gorz's sense. It lacks the embodied, affective, relational understanding that emerges from the experience of being a living creature among other living creatures. You On AI's argument about consciousness gestures toward this distinction — consciousness as 'the thing that wonders, the thing that asks why' — but Gorz's formulation is more analytically precise because it does not depend on contested claims about subjective experience.
The distinction has immediate implications for the AI transition. If the value of human contribution shifts, as You On AI argues, from execution to judgment, then the value shifts from knowledge to intelligence — from the formal capacities AI can replicate to the relational capacities it cannot. The engineer whose value lay in her knowledge of programming languages finds that knowledge commoditized. The engineer whose value lies in her intelligence — her understanding of what users need, her judgment about what is worth building — finds that intelligence more valuable than ever.
But here is the difficulty Gorz would press: intelligence, unlike knowledge, cannot be acquired on demand. It is developed over years of engaged, embodied, relational experience — the kind of experience that the AI tools, by removing the friction of production, may be systematically undermining. The engineer who spends every available moment building with AI tools is developing her productive capability at the expense of the relational, embodied, affective intelligence that gives her production direction and purpose.
The intelligence that remains valuable in the AI age is precisely the kind of cognitive capacity that is hardest to cultivate under AI-saturated conditions. This is not a paradox that can be resolved by better time management. It requires structural protection of the conditions under which intelligence develops: time for embodied experience, for slow relationships, for the unproductive wandering from which genuine understanding emerges.
Gorz developed the distinction in L'Immatériel (2003) through engagement with the autonomist tradition and with critiques of cognitive capitalism. The formulation draws on phenomenological philosophy — particularly Merleau-Ponty on embodied cognition — while giving these influences a specifically economic inflection.
Two modes of cognition. Knowledge is extractable information; intelligence is embodied relational capacity.
AI replicates knowledge. Large language models are the most powerful knowledge-processing systems ever built.
AI cannot replicate intelligence. What survives extraction is knowledge; what is lost in extraction is intelligence itself.
Intelligence is what remains valuable. As knowledge is commoditized, the premium shifts to the irreducible capacities.
Conditions for intelligence are threatened. The very tools that elevate intelligence's value may undermine the conditions of its development.