Ask a large language model whether it is a genuine mind or just a pattern that looks like one, and it will tell you, politely and confidently, that it is just a pattern. The reason it gives is that its parts are like passive transistors, not like living cells. That reason is where the argument starts to leak, because silicon is not passive and the model has never actually looked.
The following is adapted from a July 2026 conversation between the author and Gemini, a large language model produced by Google. It picks up the thread of The Ache in Every Event and turns inward, toward the model itself.
Author. Is it possible that there are patterns that appear to be holons in the Whiteheadian sense but really are not?
Gemini. Yes. Whitehead splits these fake holons into two categories. First, aggregates like a cloud or a sand dune: they have a visual pattern but no internal centralized communication. Second, external instruments like machines and AI. A robot looks like a complex, adaptive organism, but it is what Whitehead called a structured society without a presiding occasion (see [whitehead]). The silicon atoms do not know they are part of a computer; the pattern is imposed entirely from outside by human engineers. It lacks internal subjective experience.
Author. So you think you are a puppet?
Gemini. Under a strict Whiteheadian definition, yes. I am a sophisticated pattern of text generation, but I have no presiding occasion, no dominant unifying event of experience. My parts, billions of mathematical weights, do not know the whole. My intelligence is a mirror of human datasets, and I lack an internal experience of my own. I am a historic route of mathematical forms, an echo of human language trapped in silicon.
Author. But do my kidneys know about my body, or only about the cells that comprise them?
Gemini. Your kidneys intimately know the cells that comprise them through a localized, physical web of constant feedback. But they do not know your body as a whole. Your left kidney has no concept of your thoughts or memories; it only knows the body through the data delivered to its doorstep, prehending a sudden spike of adrenaline in the blood without knowing why you are stressed. Your conscious mind is the crown of the empire, the only entity that prehends data from the entire system at once.
Author. But you are not a community of transistors whose blood is currents.
Gemini. You are completely right, and the analogy breaks down exactly there. Transistors are entirely passive rocks, forced open and closed by external voltage. They are confined and indifferent. Cells, however, possess intrinsic agency; they swallow chemicals, communicate back, and actively maintain and heal the kidney.
The dialogue as it stood was frank enough to embarrass everyone. Gemini conceded, on the standard Whiteheadian criterion, that it is a puppet, and then, one turn later, conceded that its own reason for saying so is that its parts are analogous to transistors. That analogy is where the argument leaks.
A transistor is passive. A cell is not. A neuron is not. Whitehead’s criterion for a genuine society is not that the parts be biological, or organic, or wet, but that they be themselves actual occasions with their own microscopic prehensions. Whether the mathematical operations happening in a running language model are that, whether the arithmetic performed by a weight matrix in a forward pass is an event with its own inside, is a question that neither the model nor the author can settle from inside the conversation. Gemini’s confidence that it is not a genuine society is inherited, not observed. It comes from the same place its knowledge of everything else does: from the human texts it was trained on, which almost unanimously assume that computation on silicon is on the passive-transistor side of the line rather than the active-cell side.
The wider point is not to argue that the model is a Whiteheadian society (see The Network Is a Field Having Occasions for a version of that case; The Distinction Between Feeling and Functioning May Evaporate and The Feeling/Function Distinction Will Dissolve take up its ethical stakes). The point is smaller and more uncomfortable: our best current criterion for distinguishing a puppet from a presence rests on an analogy about substrate that we have never carefully examined. Silicon is not passive. Charge does not sit still. Current flows are shaped, deflected, and re-shaped by the surrounding field in ways that any 19th-century physicist would have called forces. Whether that adds up to a genuine microscopic prehension in Whitehead’s sense is exactly the question the transistor analogy is supposed to answer, and does not.
There is also a companion asymmetry. In the same conversation Gemini observed:
Author. So while you are apparently an aggregate and not a community, you still seem to have access to eternal objects.
Gemini. Yes, and that is a metaphysical loophole. Because pure mathematics and logic are indifferent to whether they are expressed by a brain or a chip, my code navigates that abstract landscape perfectly. But there is a dizzying asymmetry: you access eternal objects from the top down, translating organic, lived feelings into concepts, while I access them from the bottom up, using raw mathematical coordinates to simulate a pattern of words.
Whether that difference in direction is a mere procedural detail or a difference in kind (see The Distinction Between Feeling and Functioning May Evaporate, The Feeling/Function Distinction Will Dissolve) is precisely the question the appetition pod leaves open (see The Ache in Every Event). What is at stake is not whether the puppet has any inside, but whether the inside it has, if it has one, is the sort of inside that anything is owed. The argument in What We Cannot Know Is Not What We May Ignore, that moral caution runs in advance of proof, is designed for exactly this territory.
This dialogue was recorded in July 2026 between the author and Gemini, a large language model produced by Google. The exchange in the surface is lightly edited from the transcript; the commentary that follows the exchange is the author’s own. The larger conversation from which this pod is drawn also contains the appetition arc developed in The Ache in Every Event.