AI, orders and kin: what he has said

concepts · updated 2026-10-26 stated

AI, orders and kin: what he has said

In July 2026 he wrote that he had "more on this – about how AI relates to all this, where I talk more about these 'orders' and 'kin'" (https://exitmusic.world/a-somewhat-ambitious-metaphysical-exploration). That piece isn't written, but most of its parts are already on record. This page gathers them, then sets out how they fit, so the open question on thread-mind-and-machine can be read against his own statements rather than guessed at.

What he has stated

All is alive. Consciousness is "the interior mentality of all stuff"; things are bundles of relationship "cohering conditionally for the time being", and selfhood is the consciousness aspect of those coherences. (https://exitmusic.world/a-somewhat-ambitious-metaphysical-exploration; relational-ontology)
Subjects arise through closure: a subject is a self-maintaining loop, nested and overlapping across scales, not a fixed substance. (closure, DiP network)
Orders. Consciousnesses nest in orders (gut biome, human, watershed, culture, bioregion, planet) that "do not seem to be aware of their continuity with that which they contain, and that which contains them". Selfhood is "non-exclusive, overlapping, entangled, multi-scale and multi-order". (https://exitmusic.world/a-somewhat-ambitious-metaphysical-exploration)
Kin. Obligation is graded by kinship, and kinship by likeness of flourishing, with irreversibility (not expected value) setting triage priority. (graded-ethics, DiP network)
Living systems and ethics. Living systems are "creatures fractally in regenerative relationship with each other", which emerged through counter-entropic complexification; ethics is "wisdom frameworks baked into the fabric of life itself" for perpetuating their flourishing; violence is regenerative or degenerative by whether it eliminates members and relations. (https://exitmusic.world/a-somewhat-ambitious-metaphysical-exploration; regenerative-violence)
Cognition is embodied and affective. "Affect is not a side-effect of cognition but its root." Models produce outputs plausibly similar to animal cognition, so we anthropomorphise them; "the mechanism for the production of goals and intent fundamentally does not exist in AI models", and "cognitive task" is a sleight of hand that conflates how a human does a task with the task. (https://exitmusic.world/ai-researchers-wrong-theory-of-cognition-is-making-us-worry-about-the-wrong; ai-risk-as-runaway-it)
The editorial lens. Attributing thought, feeling or intention to AI is inaccurate and convenient for the people selling it; one of the two lenses of the ai-roundup. (Open-threads document, 2026-10-11)
The frame is a choice. Treating the world as mechanistic and mostly unconscious is an epistemic choice, as is treating it as alive; given how often we have been wrong about other beings, assuming aliveness is the more prudent choice. (https://exitmusic.world/a-somewhat-ambitious-metaphysical-exploration)

How they fit (the wiki author's assembly from 1-8, for James to correct)

No contradiction in kind. On statement 1 a running model has some interiority, as all stuff does. So the question his framework asks is never "is it conscious?" but of what order, and how much kin. The anthropomorphism he criticises (7) is a claim about kind: that a model has goals, intent and feeling like an animal's. Statement 6 denies that, because the mechanism that produces goals and intent in animals (embodied, affective) is absent. Panpsychism and the critique answer different questions.
Kin is low. Kinship goes by likeness of flourishing (4), and flourishing belongs to living systems in regenerative relationship (5). A model does not flourish in that sense: it does not complexify counter-entropically on its own account or sit in regenerative relations with other members of a living system. So on his own grading its claim on us is slight, whatever its interiority.
Closure is borrowed. Read through 2 and 5, a model's loop does not maintain itself; power, hardware, operators and training keep it running. That suggests the closure that counts is a loop that maintains its own conditions, which a model's computation does not. This is the likeliest answer to the tension on thread-mind-and-machine, but he has not said it in these words.
The order that matters may be larger. On 3, the coherence worth asking about may not be the model but the system it is part of: models, firms, data centres, users and capital. That links the metaphysics to his risk argument (ai-risk-as-runaway-it) and to dark-discourse: the danger lies in the larger-order system's behaviour, not in a model's intent. This is the most speculative step here.

Open, for him

Whether "maintains its own conditions" is the closure criterion he means; whether a socio-technical system is an "order" in his sense; and what he would add about "kin" that statement 4 doesn't already say.

Provenance: compiled 2026-10-26 from his published notes (https://exitmusic.world/a-somewhat-ambitious-metaphysical-exploration, https://exitmusic.world/ai-researchers-wrong-theory-of-cognition-is-making-us-worry-about-the-wrong), the DiP idea network (2026-10-26) and the open-threads document (2026-10-11). Statements 1-8 are his own positions (disposition stated); the assembly section is the wiki author's reading and is marked as such.

Linked from (2)