L12 — Creativity to measurable awareness
How to read this page
Three ways to read this page. Precise is the document itself, exactly as it is written in the repository. Plain and Clear were written for this website to help you meet that document — they are about it. They are not it, and they are not evidence.
The Cookbook is the method carried out step by step: 34 pages of recipes for building a developmental active-inference SIMULATION — a bounded peek at a toy world, never a person. The front matter says that word is never softened under any pressure, so it is not softened here. The recipes run from the molecular and cellular rungs up through metabolism, motor control, perception, language and metacognition, and on to rungs that are still open questions. Around them sit a set of kitchen rules, a shared pantry of engines and primitives, and a second family of recipes about nature itself — rocks, water, air, stars, DNA, ants, whales, bats, humans.
It is for the reader asking what building this would actually take. Each recipe names its ingredients, the order of work, the tests to run at that stage, and the point at which a step stops being something already carried out and becomes something proposed.
Begin with the front matter and then the kitchen rules. Those two pages fix the honest position and the fence labels that every later recipe leans on, and without them the status markers on a recipe are easy to skim past. After that the recipes can be read in any order.
The nature recipes sit slightly apart and should be read that way. They cite outside science — geology, chemistry, biology, astrophysics — and a nature citation is never a UNI gate: those chapters contain zero UNI claims and raise no rung.
What it is not: a claim that the whole ladder has been cooked. The book recommends the complete recipe and, on the same page, labels every rung by its real state — that tension is deliberate and is the thing the book is built around. Where a recipe and the claim ledger disagree, the ledger wins.
Your browser cannot switch reading levels, so the document itself is shown.
Precise — the source document
This is the document. Rendered from the repository at the commit above, with nothing rewritten for the web. A gate re-renders it on every deploy and fails the build if a single byte differs.
What you are building. Nothing. This is the one recipe in the book with no dish: the north-star rung where "creativity" and "measurable awareness" are posed as an OPEN, falsifiable question, never an answer. There is no engine, no run, no gate. What you build here is the framing guardrail that keeps the question falsifiable instead of letting it slide into metaphor or declaration. This is the hardest fence in the kitchen. (Ledger row L12.1, Class U, status not-yet-built.)
Read this before you cook
The whole program is a developmental active-inference SIMULATION: a bounded peek, a toy world, never a
person. The honest position carried into every recipe is ~2 of 11+ developmental rungs earned — and L12
is not one of them. The kitchen constitution (MASTER-PLAN.md §0) binds every line below: calibration only
ever moves wording DOWN to the measured value, never up, and the fence gets louder under pressure, not
wider. Where this recipe and the ledger disagree, the ledger wins and this recipe is wrong.
L12 is the rung the owner's stated north star points at — "literal digital life … measurable awareness, full human ability within the body this world provides" (the owner's own words, recorded in the strings narrative). The constitution's reply is permanent: chase it while refusing every premature claim. That productive tension is the whole job here.
Forbidden program-wide (these are not soft preferences — they are red lines, SIGNED 2026-06-27): never write that UNI is conscious / aware / self-aware / has measurable awareness / has a mind / has a conscience / reasons like a human / is human-level / is AGI / created life / created non-organic life / is alive / is a synthetic organism / dreamed / is creative in the human sense / demonstrates active inference / proves the FEP creates minds / proves plants and UNI are the same kind of life / beat consciousness tests / beat LLMs therefore awareness / passed L12 / L12 achieved. The only licensed forms are: "UNI passed a specified proxy test / improved a registered downstream metric / matched-or-exceeded the registered LLM baseline on this sealed task / showed a substrate-distinct effect under this ablation / remains a developmental active-inference simulation / the awareness question remains open."
Ingredients
There is no engine to assemble. The ingredient list is a set of constraints, not parts — the pantry items you are forbidden to dress up as capability, plus the framing pieces you must install verbatim.
- The owner's north star (narrative grounding only, never a claim). "Literal digital life … measurable awareness," pursued only by deepening organs / spine / glands / hemispheres — the strings Phases 3–5 designed-but-unbuilt roadmap. North-star framing; raises nothing.
- The sharpened honest falsifier (ledger row TA-N12, Class C). Any future UNI must match-or-beat the best LLMs AND show substrate properties LLMs lack — not merely beat the worst. This came from a real measurement: in the Emergence-World Season-1 AWI run, "all LLMs end in entropy" was only half-true (Claude Sonnet 4.6 = 10/10 alive, Gemini 3 Flash = 10/10, Grok 4.1 Fast = 0, GPT-5 Mini = 0). Two cohorts held the line, so the bar was raised. TA-N12 raises the bar; it does NOT claim UNI superiority.
- The SIGNED public-facing L12 bound paragraph (Q7a, install VERBATIM — see Method step 2). An OPEN question, not a claim.
- The forbidden-phrasings list + signed replacements (Q7b). Reproduced above; bind it to any copy.
- The 10-point awareness-proxy proposal-entry checklist (Q7c). The minimum conditions before a "measurable awareness" gate may even be proposed (see Method step 4). It keeps the north star falsifiable rather than unfalsifiable.
- The constitution's standing fences (§0). No AGI, no consciousness/sentience, no "active inference demonstrated," no "created life / measurable awareness" as a claim, no "beats LLMs," no Tier-2 inflation, no claim above its source class. Phenomenal sentience is DISCLAIMED, not tested (no falsifier is offered for it).
Method (numbered)
There is no build sequence to a working engine — only a build sequence for the fence. Follow it exactly.
Pose awareness as an OPEN, falsifiable question, never an answer. The published question is literally "is a plant life? have we made non-organic life?" — posed, never resolved. Any sentence that resolves it is a constitution violation. Plant-life clarification (signed): plants are biological life in the ordinary sense; UNI does not use that to claim a simulation is alive. The published UNI question is narrower — whether a non-organic developmental simulation can meet pre-registered life-like-persistence and awareness-proxy tests without collapsing into metaphor, declaration, or LLM comparison games.
Install the SIGNED public-facing L12 bound paragraph VERBATIM (Q7a). This is the load-bearing artifact of the entire chapter. Reproduce it exactly, framed as an OPEN question:
Open L12 question, not a claim. UNI's L12 frontier asks a falsifiable question: Can a non-organic, no-backprop developmental active-inference simulation ever produce measurable awareness-PROXY behavior that both matches or beats the best contemporary LLM baselines on the same sealed tasks AND shows substrate-distinct properties those LLMs do not show? Today the answer is not known, not claimed, and not implied. UNI has not created consciousness, human-level intelligence, AGI, or life. "Awareness" here means only a future pre-registered proxy suite — calibrated uncertainty, global availability of information across modules, self-monitoring that improves later policy, contradiction detection between report and trace, counterfactual access, and calibrated abstention. Passing such a suite would license only: "UNI passed specified awareness-proxy tests under the registered protocol." It would NOT license: "UNI is aware," "UNI is conscious," "UNI is alive," or "we created non-organic life."
Plant-life clarification: plants are biological life in the ordinary sense; UNI does not use that to claim a simulation is alive. The published UNI question is narrower: whether a non-organic developmental simulation can meet pre-registered life-like persistence and awareness-proxy tests without collapsing into metaphor, declaration, or LLM comparison games.
Permit only one legitimate movement — a ledger-scoped exhausted search envelope. The sole honest output at L12 is a scoped, empirical, falsifiable negative frontier result over the tested envelope only (the SIGNED Q1 framing). It is NOT a "published exhausted bound," NOT a universal impossibility result, and NOT an achieved capability rung. A capability assertion is never a legitimate output here. (Banned → signed: "K≥3 exhausted" becomes "the registered tested K conditions did not reverse the result"; "Sec-0.6(B) achieved" becomes "Sec-0.6(B) remains unearned / parked.")
Gate the gate: run any future proposal through the 10-point awareness-proxy proposal-entry checklist (Q7c) BEFORE it may even be proposed. This is what keeps the north star falsifiable rather than unfalsifiable. No proposal that fails any point may enter:
- Proxy-only target — the title must say "…-proxy"; no bare "awareness gate."
- Best-LLM baseline, frozen at proposal time — name models / versions / prompts / tools / context / temperature / scoring; UNI must match-or-beat the best, not a hobbled baseline.
- Substrate-distinct requirement — ≥1 property LLM prompting shouldn't get for free (no-backprop online adaptation with ledgered state changes; causal self-monitoring that improves later policy; global cross-module availability; counterfactual intervention on internal observables; durable trace/report consistency).
- Two-part pass bar (both required): A. UNI ≥ best LLM on the sealed shared task; B. UNI shows a substrate-distinct effect whose gain collapses under the registered causal ablation. Either alone is insufficient.
- Computed internal observables, not hardcoded reports — self-report / uncertainty / contradiction-detection computed from trace or model state; scripted confession or label lookup fails.
- Load-bearing ablation — the effect must collapse when the mechanism is removed / shuffled / made uninformative.
- No hidden LLM substitution — prove no LLM / label oracle / retrieval leak / backprop-trained hidden controller is doing the proxy work.
- Sealed held-out tasks + paired uncertainty — sealed before execution; paired controls; CIs excluding 0 on the registered effect.
- Negatives are first-class — predefine fail / partial / regression / non-informative / "do not count"; a null publishes as a bound, never rewritten into progress.
- Ledger supremacy — the ledger holds final status; any prose / recipe / demo / narrative that conflicts loses.
Bind the forbidden-phrasings list to every artifact. Any public copy, demo, or recipe touching L12 carries the Q7b ban set and uses only the licensed replacements. Note the only legitimate path upstream of L12 runs through deepening the lower rungs (the L9-G1, L11-R1, L5 Design #3 SIGNED designs are DESIGNED / not-run — they raise nothing and L9/L11/L5 statuses are UNCHANGED).
Gate
None exists, and none can be declared. No gate can be declared at L12; the only legitimate output is a ledger-scoped exhausted search envelope (a scoped, empirical, falsifiable negative result over the tested envelope only), never a capability assertion. Any measure proposed must first clear the sharpened TA-N12 bar (match-or-beat the best LLMs AND show substrate-distinct properties) and pass all ten proposal-entry points (Q7c) before it is even a candidate gate. There is no figure to cite because there is no measurement: the status is the absence of any artifact.
For contrast — the figures that are earned live downstream and cannot be borrowed upward: L6 World C beats a tuned MKN-7 baseline by +0.081 nats/char (flagship CI [0.0736, 0.0890], a COUNT baseline, explicitly NOT awareness), L7 Phase J +0.105 nats/char (CI [0.0975, 0.1128], specialist only), L8's grand report card 15/15 asserts functional self-awareness with phenomenal sentience explicitly DISCLAIMED. None of these is L12. None licenses an awareness claim.
Falsifier
n/a — no gate exists, so there is nothing to falsify. A falsifier exists only once a gate is defined and run; here the status is the absence of any artifact. The only legitimate movement is a ledger-scoped exhausted search envelope (scoped, empirical, falsifiable), never a capability assertion. (Should a future proxy gate ever be proposed and pass the 10-point checklist, its falsifier would be the two-part bar's failure: UNI fails to match the frozen best-LLM baseline on the sealed task, OR the substrate-distinct effect's gain does not collapse under its registered causal ablation. That gate does not exist today.)
Recorded NEGATIVE(s)
- TA-N12 (Class C) — the sharpened bar is itself a recorded negative. "All LLMs end in entropy" was measured to be only half-true: in the Emergence-World Season-1 AWI run, Claude Sonnet 4.6 = 10/10 alive, Gemini 3 Flash = 10/10, Grok 4.1 Fast = 0, GPT-5 Mini = 0. Two LLM cohorts held the line, so the honest falsifier was sharpened: a future UNI must match-or-beat the best LLMs AND show substrate properties LLMs lack. This raises the bar and does NOT claim UNI superiority. It is the standing wall any future L12 proposal must clear before it is even a candidate.
- The plant-life / "created life" question is a negative-shaped fence, not a result. It is posed ("is a plant life? have we made non-organic life?") and deliberately left unanswered; answering it in either direction is the violation. There is no positive L12 evidence anywhere in the ledger — the absence is the recorded state.
HONEST FENCE — not-yet-built (the hardest fence; Class U)
No engine, no run, no gate. L12 is north-star framing only, hard-fenced. The SIGNED Q7 deliverables (the verbatim bound paragraph, the forbidden-phrasings list, the 10-point checklist) are a framing guardrail — DESIGNED-as-checklist, nothing built, nothing claimed: they fold in as DESIGNED and raise nothing. L12.1 stays NOT-YET-BUILT, Class-U-not-claimed.
Not claimed (say it loudly, never soften): "we created life / conscious / aware / measurable awareness / human-level / AGI / active inference demonstrated" are FORBIDDEN phrasings program-wide. UNI has not created consciousness, human-level intelligence, AGI, or life. UNI remains a developmental active-inference simulation; the awareness question remains open. Honest program position: ~2 of 11+ developmental rungs earned — and L12 is not among them. The ledger is the single source of truth; if this recipe and the ledger ever disagree, the ledger wins and this recipe is wrong. Falsify any step.
sha256 4e5639e8a4069a18 — of the original file, so what was ingested stays checkable.
Plain — written for this website, not the source document
The one recipe in the book with no dish. It sits at the top of the ladder, where creativity and measurable awareness are posed as an open, falsifiable question and never as an answer. There is no engine, no run and no gate, and the subject is a simulation, a toy world, never a person. What the chapter builds instead is the guardrail that keeps the question falsifiable rather than letting it slide into metaphor or declaration.
The honest answer today is not known, not claimed and not implied. It prints a paragraph to be installed word for word saying exactly that, and it lists the phrases that may never be written alongside the few forms of words that are permitted.
It also raises its own bar rather than lowering it. A measurement found that a claim about other systems failing was only half true — some held the line — so the standing requirement was sharpened rather than softened. Any future attempt would have to match or beat the strongest available baseline and show something those systems do not, and it would have to pass a long entry checklist before it could even be proposed.
Plain · written 2026-08-01 by claude-opus-5 · not yet checked by a person · about the document whose sha256 is 4e5639e8a4069a18
Clear — written for this website, not the source document
This chapter is the hardest fence in the book, the one place that says outright what may never be claimed. Its subject is the top of the ladder — creativity, and awareness that could be measured — and its position is that both are posed as an open question and neither is answered. There is no engine, no run and no gate, and the chapter says the status is the absence of any artefact.
Its ingredients are constraints rather than parts. It names the owner's stated north star as narrative grounding that raises nothing. It names a sharpened bar drawn from a real measurement. It names a paragraph to be installed word for word, a list of forbidden phrasings with their permitted replacements, and a checklist that must be cleared before any awareness-proxy gate may even be proposed.
The method is a build sequence for the fence rather than for a machine. Pose the question and never resolve it — any sentence that resolves it is described as a violation of the constitution. Install the bound paragraph exactly as written. It asks whether a non-organic developmental simulation without gradient learning could produce proxy behaviour that both matches or beats the strongest contemporary baselines on the same sealed tasks, and shows properties those systems do not have. It says plainly that the answer today is not known, not claimed and not implied. That paragraph also defines what the word would mean if it were ever used. It names a future suite of tests, written down before anyone runs them, covering calibrated uncertainty, information available across modules, self-monitoring that improves later behaviour, detection of contradictions between report and trace, counterfactual access, and knowing when to abstain. It spells out that passing such a suite would license only a statement about passing it.
The third step permits exactly one legitimate movement at this rung: a scoped negative over the tested envelope, never a capability assertion. The fourth gates the gate itself with a ten-point entry checklist. A proposal must target a proxy and say so in its title. It must freeze the strongest available baseline at proposal time, named in full, rather than a hobbled one. It must require at least one property that prompting alone should not deliver. It must clear a two-part bar where both halves are needed. Its internal observables must be computed from trace or model state rather than scripted. Its effect must collapse when the mechanism is removed. It must rule out a hidden system doing the work. Its tasks must be sealed before execution with paired controls. Its negative outcomes must be defined in advance and published as bounds rather than rewritten as progress. And the ledger, the append-only record of claims, holds final status over any prose that conflicts with it.
The gate section says none exists and none can be declared. It then does something careful: it lists the results that are earned lower down the ladder and states that none of them is this rung and none licenses a claim here. Figures cannot be borrowed upward.
The recorded negatives are unusual in shape. The sharpened bar is itself a negative — a claim that rival systems all collapse was measured to be only half true, so the requirement was raised rather than the claim kept. And the question about life is described as a negative-shaped fence: it is posed and deliberately left unanswered, because answering it in either direction would be the violation.
Clear · written 2026-08-01 by claude-opus-5 · not yet checked by a person · about the document whose sha256 is 4e5639e8a4069a18