PhronesisThe easy line
Start by granting the first seed, plainly. Synthetic thought is ai.
Synthetic and artificial are the same word wearing two coats. Artificial comes from the Latin artificium — a work of skill, a thing made by craft (ars, skill, plus facere, to make). Synthetic comes from the Greek syntithenai, to put together — syn, together, plus tithenai, to place. Both mean made: assembled by hand and intention rather than grown. And thought and intelligence overlap so far that the gap barely holds a claim. So "synthetic thought is ai" reduces to "made thought is made intelligence." That is close to a tautology.
Herbert Simon fixed the sense the word carries. The sciences of the artificial, he wrote, are "concerned not with the necessary but with the contingent — not with how things are but with how they might be — in short, with design." The artificial is the made, the designed, the intended, set against the natural — the grown, the necessary, the given. That is the line built into the word.
An honest note names a tautology as a tautology. The first seed is a clean, compressed line, and it is true, and it is mostly a definition being unpacked. It is not a discovery, and dressing a definition as a finding is the one move a working note should never make.
Where the second line breaks
The second seed is the reason to write anything at all, because as stated it is false. Any cognized practice is ai — no.
A potter centering clay is running a cognized practice: perception, correction, timing, a whole trained loop of hand and eye. A player mid-game is running a cognized practice. Neither is artificial intelligence. Each is the original kind — the grown, biological kind, the very thing the artificial is defined against. If a practice is cognized simply by being done well by a living practitioner, then it is exactly not made-not-grown, and the word "artificial" has nothing to bite on.
That is the cost of the over-reach. "Artificial" carries the made-vs-grown line, and the second seed erases it — it calls every act of cognition artificial, which drains the word to nothing. If all cognition is ai, the term marks no boundary and says no thing. A word that includes everything it might exclude has stopped working.
Hubert Dreyfus pressed the same seam from the side of skill. Expert practice does not present itself as rule-following: the practitioner mid-flow is coping with a situation, not consulting a table of instructions they could recite. (His stronger claim — that machines can therefore never match human skill — is contested, and this note does not lean on it. The narrow point is enough: living skilled practice is not, on its face, a rule being executed.) So the living practitioner is the clearest counter-example the second seed has. Cognized, yes. Artificial, no.
The threshold
The seed points at something real, though, and it can be repaired into a sharper claim. Here is the corrected line, offered as a proposal.
A practice becomes artificial at the threshold where it is formalized enough to run without the practitioner — lifted out of the one who does it and set down into a rule, a procedure, a machine that executes on its own. The hinge is the effective procedure: a method specified completely enough that no understander is needed to carry it out. Turing made this exact. A computation reduced to fully specified steps runs on a simple machine, and the machine does not know what it computes. That is cognition externalized — pulled out of the cognizer and made to execute.
So the artificial is not cognition. It is cognition made portable in the strongest sense: made to run somewhere other than the mind that first held it. The extended-mind argument is a near cousin here — Clark and Chalmers show cognition can reach out into notebook and world — but it is adjacent, not identical, and the note keeps it honest: they describe cognition extended with the agent still in the loop, which is a step short of cognition run without them. The threshold this note names is the second thing: the loop closed around the procedure, the practitioner gone.
Cognition made to run outside the one who cognizes is the artificial — cognition alone is not.
The limit
The threshold has a far wall, and it must be in the note or the correction over-reaches in its own direction.
Some practice resists being lifted out at all. A competence can be used and still cannot be fully told — Polanyi's line, that "we can know more than we can tell" (the popular form drops the "can"). The craftsman's hands, the read of a face, the balance on a bike: real, structured, and resistant to being written down completely enough to run without the body that has it. Tacit knowledge develops this — the grammar that never rose into words. Where the tacit holds, the lifting-out fails, and the practice stays with the practitioner. It cannot be made artificial because it cannot be made to run alone.
Un-shortcuttability marks the same wall from the other side: some outcomes have a knowable rule and still cannot be substituted for the happening — the only way to the result is to let the thing run. The artificial threshold is where a practice can be closed into an effective procedure; the tacit and the un-shortcuttable are where it cannot. Together they draw the boundary of the artificial. It is a real region, not the whole of cognition.
The site's own instance
This move is not only argued here; it is enacted on the site. The displacement framework is cognition lifted out and set to execute — a practice of reading a state, its ground, its cost of staying, its accumulation, formalized into a single sheet that runs the same whether or not its author is in the room. That is the threshold met in the open: the framework is displacement-cognition externalized into a specified procedure. Cognitive displacement names the machinery this externalization runs on. So the note's own claim has a working example one click away — cognition made to run, and the honest limits of that same move sitting right beside it.
Keep the correction where it can be seen, then, and do not bury it under the tidy first line. The easy seed is a definition owning up to being one. The real seed, fixed, is the threshold: a practice becomes artificial not by being cognition but by being made to run without the one who cognizes — and only up to the wall where the tacit and the un-shortcuttable refuse to be lifted out at all.
Kin to Tacit Knowledge and Un-shortcuttability — the two limits of the lifting-out — with Cognitive Displacement for the mind's own machinery, and a light see-also to externalism.
Rests on: Simon, The Sciences of the Artificial (1969), for what "artificial" means — the made and designed, set against the natural; the etymologies of artificial (Latin artificium, made by skill) and synthetic (Greek syntithenai, to put together); Turing, "On Computable Numbers" (1936), for the effective procedure — a method that runs without an understander; Polanyi, The Tacit Dimension (1966), for "we can know more than we can tell," the limit of formalization; Dreyfus, What Computers Still Can't Do (1972/1992), for expert skill as coping rather than rule-following — his stronger no-machine-can-match claim is contested and not leaned on here; and Clark & Chalmers, "The Extended Mind" (1998), as a light cousin on externalization — cognition extended with the agent, which is a step short of run without them. The threshold claim — that a practice becomes artificial exactly where it is formalized enough to run without the practitioner — is a proposal, offered to be argued with, not a proven identity.