Part X — On Naming as the Last Step
DEFINITIONS
I. By thing I understand any new technology, or any new condition of life, that has come into existence.
II. By deployment I understand the extension of the thing into the world.
III. By naming I understand the moment at which the general public has a stable, agreed-upon word for the thing.
IV. By placeholder I understand a word treated as stable while pointing at a moving target.
V. By loud language I understand decisive slogans, confident predictions, urgent acronyms.
VI. By signal I understand what the work actually achieves.
AXIOMS
I. The thing exists first.
II. The deployment happens second.
III. The naming comes last — often by a lot.
IV. If a name can be cleanly given, the deployment is already mature.
V. The loudest people in a new era are typically the ones who understand it least.
PROPOSITIONS
Prop. I. Every prior technological transition shows the same order: thing → deployment → name. Proof. The first commercial transistor went on sale in 1954, sold for hearing aids; the word transistor entered general English usage only after the device had been embedded in millions of products for over a decade. Electricity was generated commercially for thirty years before electron entered the language. Computers existed fifteen years before software became an ordinary word. The internet was used by millions before email was unhyphenated. The cloud took most of a decade to become unremarkable. Q.E.D.
Prop. II. The general public knows the new condition before it has a word for it. Proof. Buyers of transistor radios knew the new radios were smaller and lighter than the old ones. They did not know why. The why required a word. The word came later. Q.E.D.
Prop. III. If a name can be cleanly given, the deployment is already mature. Proof. By Axiom IV. Mature deployments permit stable language because their shape has settled. Settlement permits naming. Q.E.D.
Corollary. If you can name something cleanly and confidently, you are usually too late to be early to it.
Prop. IV. If a name cannot be cleanly given, the thing is still being seen, not yet named. Proof. Contrapositive of Proposition III. Instability of language reflects instability of the underlying thing. Q.E.D.
Prop. V. We are in this phase now, for AI. Proof. "Artificial intelligence" is a 1956 marketing term coined for a Dartmouth grant proposal; it has never stopped being slightly misleading. "Machine learning" describes one technique among many and treats the result as an artifact of the technique. "Large language model" is technically accurate and conveys nothing about what the thing does. "Generative AI" is vague enough to cover everything from chatbots to image generators to music systems, which is also why it tells the listener almost nothing. None of these is stable; all are placeholders. Q.E.D.
Prop. VI. "AGI", "agents", and "alignment" are placeholders treated as stable concepts. Proof. They are debated as if each pointed at one identifiable kind of thing. None does. Each is contested, drifting, used differently by different speakers in the same conversation. The discourse pretends to know what it is debating, because pretending is more comfortable than admitting one is naming a moving target. Q.E.D.
Scholium. — People argue about whether AGI is close, as if AGI were a stable concept. People argue about whether agents are real, as if the word pointed to one kind of thing. People argue about whether alignment is solvable, treating it as a problem with crisp boundaries. None of these names is stable. They are placeholders, and the work of pretending they are not is the daily work of an industry that has confused volume with clarity.
Prop. VII. The real names of the present era — the ones in unremarkable use forty years from now — have probably not been coined yet. Proof. By Proposition I and the historical pattern. They will be coined by someone not famous, in a context not currently being watched. They will spread quietly. By the time they are established, the underlying technology will have settled into a shape today's engineers would not fully recognise. Q.E.D.
Prop. VIII. Loud language is a sign of an industry talking about itself, not understanding itself. Proof. By Axiom V. The people who actually understood electricity in 1900 were not the ones writing manifestos. They were the ones quietly running cables. Q.E.D.
Prop. IX. The discipline that follows is: notice when the language is loud. Proof. Loudness is the early warning. By Proposition VIII, it marks the place where attention is, not where understanding is. Q.E.D.
Prop. X. The Work, or, the Signal. Proof. By Proposition VIII and Proposition IX. Where the language is loud, the signal is in the work; the names will come last; the work is now. Q.E.D.
Corollary. The noise is just noise.
Final Scholium. The people who will turn out to have been right about this era are mostly not the ones who are loudest right now. They are the ones who can sit with the fact that we do not yet have the right words, and who keep working anyway, knowing that the names will come last.
When the names come, you'll know the moment is over. The new normal will already have arrived.
Until then:
The work, or, the signal.
The noise is just noise. The signal is in the work.