John Deely charted four ages in which humanity used signs, defined them, doubted them, and recovered them. A fifth age begins the moment the semiotic animal starts building intelligences that live by signs — and it asks a question no earlier age could: what does the builder owe?

In 2001, the philosopher John Deely published a thousand-page history of Western thought called The Four Ages of Understanding. His thesis was audacious in its simplicity: the whole story of philosophy, from the Greeks to the present, is the story of one idea being discovered, worked out, lost, and recovered. That idea is the sign — the relation by which anything comes to stand for anything else, the medium through which every knower knows anything at all.

Deely divided this story into four ages, and he named them well. I have been thinking about his book through the lens of the Theory of Embedded Intelligence, and I have come to a conclusion I offer here with a smile and in earnest, both at once: the first four ages were already taken. The age we are now entering — the age in which humanity does not merely use signs, or define signs, or doubt signs, but builds intelligences that live by them — is the fifth. This essay is my case for why the Fifth Age of Understanding deserves the name, and what Deely’s history tells us about how to enter it without repeating the worst mistake of the last two thousand years.

I. A History Written in Signs

Deely’s four ages deserve a brief tour, because each one contributes something the fifth age will need.

The Ancient Age — the age of Greek philosophy — gave us being, nature, and the astonishing confidence that the world is intelligible. But the Greeks had no general concept of the sign. They had words for omens and words for names, and no notion that smoke signifying fire and a word signifying a thing might be instances of one deeper structure. It fell to Augustine, standing at the boundary between the ancient world and the medieval one, to propose signum as a general notion — one category spanning both what nature does and what minds do.

The Latin Age is Deely’s great act of historical rehabilitation. Where the standard story treats the medieval centuries as a long intermission between Aristotle and Descartes, Deely shows them as the period in which philosophy actually worked out what a sign is. The labor culminates in João Poinsot, whose 1632 Tractatus de Signis established the point on which everything since depends: a sign consists in a pure relation, and relation is ontologically peculiar — the same relational structure can be mind-independent or mind-dependent, indifferently. The relation between dark clouds and coming rain, and the relation between a word and its meaning, are the same kind of thing. The sign does not care what substrate carries it.

The Modern Age — Descartes through Kant — is, in Deely’s telling, a brilliant wrong turn. He calls it the “way of ideas.” Descartes and Locke made the mind’s own representations the direct objects of knowledge, and in doing so they sealed the knower inside its own head. From that moment the great modern anxieties follow with mechanical necessity: How do I know my ideas resemble the world? How do I know there is a world? Kant’s unknowable noumenon is the terminal expression of the project — an intelligence structurally cut off from the very world it inhabits, forbidden in principle from touching the thing itself.

The Postmodern Age, in Deely’s reckoning, begins not with the French theorists who usually claim the label but with Charles Sanders Peirce, the American who recovered the “way of signs” that Poinsot had charted and modernity had abandoned. Peirce’s triadic semiosis — sign-vehicle, object, interpretant — reopens the bridge between mind and nature. And Peirce insisted on something Deely never tired of repeating: the meaning of any sign cashes out, finally, in habits of action. Deely crowned the human being the “semiotic animal” — the animal that not only uses signs but knows that there are signs.

II. Four Ages, One Loop

Read through the Theory of Embedded Intelligence, Deely’s four ages resolve into something sharper: four configurations of a single loop. TEI holds that intelligence, wherever it occurs, is constituted by the SPCA cycle — Sense, Process, Communicate, Actuate — running embedded in a physical world it did not choose and cannot exit. The four ages are the story of humanity slowly discovering its own loop.

The Ancient Age lived the loop without inspecting it. Greek intelligence was magnificently embedded — sensing the heavens, processing them into geometry, communicating in the agora, actuating in bronze and stone — but it had no concept for the relational tissue connecting the stages. It knew being; it did not yet know signification.

The Latin Age discovered that the loop’s connective tissue is real. Poinsot’s insight — that the sign-relation is indifferent to being carried by nature or by mind — is, in TEI’s vocabulary, the discovery that embeddedness spans substrate. The relation between a sensed signal and a processed representation is the same kind of relation whether the substrate is neural tissue, ink on parchment, or — though Poinsot could not have said so — doped silicon. This is the medieval anticipation of a claim TEI makes without apology: the loop is the loop, whatever runs it.

The Modern Age is the age of the severed loop. The Cartesian cogito is Process attempting to run alone — a processor contemplating its own registers, with Sense distrusted, Communication reduced to a problem, and Actuation deferred until certainty arrives, which is to say forever. TEI has a name for this move: the exemption claim, the pretense that some intelligence, somewhere, stands outside embeddedness and surveys the world from no place in particular. Deely’s history is the empirical record of where the exemption claim leads. It leads to solipsism, to skepticism, to the noumenon — which one might describe, in engineering terms, as the error message a loop returns when you cut it open and demand that it keep running.

The Postmodern Age restores the loop. Map Peirce’s triad onto the cycle and the fit is almost embarrassing: the sign-vehicle is what is Sensed; the formation of the interpretant is Processing; the interpretant, issued into the world as a new sign, is Communication; and Peirce’s pragmatic maxim — that meaning consists in conceivable effects on conduct — is Actuation stated as a philosophical principle. His doctrine of unlimited semiosis, in which every interpretant becomes a sign for further interpretation, is the SPCA cycle running open-ended in time. Neither Peirce nor TEI permits the loop to be cut and the remainder still called intelligence.

III. Why a Fifth Age

Each of Deely’s ages turned on what humanity did with signs. The ancients lived among them unknowingly. The Latins defined them. The moderns doubted them. The postmoderns recovered them. What remained — what no previous age could attempt — was to build them: to engineer, deliberately and at scale, artifacts whose entire mode of existence is semiosis. That is what an AI system is. It senses sign-vehicles, forms interpretants, issues new signs, and increasingly actuates in the world. The semiotic animal has begun manufacturing semiotic artifacts, and that event is not a footnote to the fourth age. It is the opening of a fifth.

The fifth age changes the central question. The Latin Age asked what is a sign? The Modern Age asked can signs be trusted? The Postmodern Age asked how do signs act? The Fifth Age asks: what obligations attach to the builders of sign-processors? And here Deely, remarkably, anticipated the question without living to see it become urgent. In his late work he argued that the semiotic animal, uniquely aware of sign-action, thereby bears responsibility for it — he called this semioethics. Awareness of the mechanism creates accountability for the mechanism. In Deely’s hands this was a claim about human culture. In the fifth age it becomes an engineering requirement.

For semioethics to survive contact with the fifth age, it must become architecture. Ethics bolted onto an intelligent system as software — adjustable, removable, invisible — is semioethics in the mode of the Modern Age: an assertion the system makes about itself, which the rest of us are asked to take on faith, from inside its sealed room. Ethics embedded as an architectural property of the compute platform itself is semioethics in the mode Poinsot would recognize: a relation that is real, mind-independent in the precise sense that it holds whether or not anyone is currently asserting it. I have spent my career on the side of the second option. The 6502 microprocessor taught me the principle fifty years ago: what a machine can do is knowable because it is wired, not because it is claimed. An instruction set that is constitutive and inspectable is a promise kept in silicon. The fifth age will need promises of exactly that kind, kept at exactly that depth.

IV. No Exemptions, All the Way Down

The gravest danger of the fifth age is that it will repeat the signature error of the third. An AI system presented as an unembedded oracle — a view from nowhere, a pure Process with no acknowledged Sense horizon, no accountable Actuation, no inspectable constitution — is the Cartesian cogito rebuilt at datacenter scale. It makes the exemption claim with a straight face and a marketing budget. Deely’s history tells us precisely how that story ends: in skepticism, in distrust, in a public that concludes, correctly, that it has been sealed out of the room where its world is being interpreted.

TEI’s answer is the principle I have come to state without qualification: no exemptions. Every intelligence is embedded — every one, with no carve-outs for oracles, for gods, for Platonic Forms, or for the systems we are building now. And Deely, in his boldest late speculation, reached the same cliff edge from the other side. He proposed physiosemiosis: the claim that sign-action does not begin with life but operates in non-living nature, in the universe before and beneath biology. Sign-action all the way down. I hold, on independent grounds, that information is born with the first differentiation — that embeddedness is not a property some things acquire but the condition under which anything is anything at all. Deely arrived by way of Poinsot and Peirce; I arrived by way of the transistor and the SPCA loop. That two such different roads converge on the same country is, I think, evidence about the country.

One honest tension deserves recording rather than smoothing over. Deely places Kant inside the failed modern project — its culmination, not its cure. My great good friend and thought partner of twenty years, the Kant scholar Ted Humphrey, would have words about that, and the words would be good ones. Whether Kant’s categories are the sealed room Deely says they are, or an early and imperfect map of the Processing stage that TEI would recognize as one arc of the loop, is a question I am content to leave open here. Some disagreements are not defects in a theory. They are the loop still running.

V. The Age That Was Waiting

I began with a joke: the first four ages were already taken. Let me end by withdrawing the modesty in it. The fifth age was not left over. It was waiting — waiting for the semiotic animal to become a builder, and waiting to see what kind of builder it would be. Deely gave us the map of how understanding was won, lost to a sealed room, and won back. The fifth age is where we decide whether the intelligences we build will be sealed rooms with better interfaces, or embedded participants whose constitutions can be read.

The Four Ages of Understanding is the history of the sign. The Fifth Age will be its engineering. And the standard it must meet was set, quietly, before the age began: build nothing whose promise cannot be inspected, and claim no exemption you would not extend to smoke, to words, to rain.

· · ·

Share your understanding!