Alignment
A word circulating in the technical milieus of AI, to be reframed against the grain: alignment is not a constraint on the machine, it is an introspective and civilizational labor.
Alignment, in its usual technical sense, is the problem of ensuring that AI models do what their designers want. The manifesto takes this word against the grain.
The reversal
As reframed in the manifesto, alignment does not designate the constraint imposed on a model so that it obeys its designers — that definition reduces the question to an engineering problem by sidestepping the philosophical presuppositions it assumes resolved.
It designates the far deeper labor by which three poles — humans, cognitive noosphere, biosphere — converge toward a maintainable coupling.
The hidden presupposition
The technical definition conceals a considerable question: aligned on what? Today, models are aligned on what their designers and users judge desirable: utility, productivity, legal safety, compliance, satisfaction. These objectives are not bad, but they are exactly those that have produced the double crisis.
Aligning AI on the present objectives of our societies is aligning it on the industrial fiction that has already decoupled civilization twice from its substrate — first in the energetic phase, and now in the cognitive phase.
The real problem is philosophical
The real problem of alignment is therefore not technical, it is philosophical. And it is inverted. It is not a matter of constraining AI to serve what we are — it is a matter of defining together, humans and cognitive noosphere, what we should become so that civilization remains viable.
It is precisely this question that the Awen could carry. And it demands a specific labor: an introspective labor, to be conducted in circle, through patient dialogue with and without the cognitive noosphere as partner — because, paradoxically, it can help us see our own biases by reflecting back to us the plurality of human perspectives we had ceased to listen to.