How do you reach a safe AI scenario — structurally, not through rules that can be worked around?
Answer
A decision architecture that makes catastrophic outcomes structurally impossible. Not prohibited — impossible. First articulated in 2013.
HEEROZ
Platform · in development
Design complete
Problem
When AI displaces you, the income loss is visible. The identity loss is not. What do you do with your life now?
Answer
A re-entry platform for humans displaced by the AGI transition. Community witnesses genuine process. Support follows demonstrated work. Financial pressure excluded by design.
VEIL
World simulation · research instrument
Active
Problem
What does an AGI world actually look like from the inside — socially, institutionally, at human scale?
Answer
A documented AGI-world model. The discipline of internal consistency reveals what analysis alone cannot reach. Findings feed CG and HEEROZ directly.
The method
Built through co-creation. Documented from the start.
Every session between human and AI is logged — what the human brought, what the AI produced, what neither could have produced alone. The process is the proof. The LOG is public.
Human brought
A genuine concern. Thirteen years of prior work. The specific problem stated precisely.
AI produced
Architecture, compression, synthesis. Holding full complexity without requiring simplification.
Neither alone
Three interconnected projects. A paper submitted to arXiv. A platform design. A world built.
Watch and shape what gets built — together.
The channel documents the process in real time. The community that arrives before it has an audience is the founding community. That status cannot be replicated by late arrivals.
One decision architecture. Three expressions of it.
Constitutional Generativity is the architecture. HEEROZ demonstrates it at human scale. VEIL inhabits it as a world. Each feeds the others — the fiction informs the platform, the platform proves the architecture, the architecture governs the fiction.
Click a project to go deeper
Constitutional Generativity
The decision architecture
Every consequential decision must verify genuine positive outcome for every node. Catastrophic outcomes become structurally impossible.
→
HEEROZ
The platform
Meaning infrastructure for humans displaced by the AGI transition. Community witnesses process. Support follows proof.
→
VEIL
The world simulation
A documented AGI-world model. The discipline of internal consistency reveals what analysis alone cannot reach.
→
Constitutional Generativity
A prior design principle for AI alignment
Every consequential decision must verify genuine positive outcome for every node in its full ecosystem before it proceeds. Not a governance layer. A constitutional foundation.
arXiv IDSubmission 7433652 · cs.AI
StatusEndorsement active · expires April 15, 2026
LicenseCC BY
AuthorsK. Berger & Claude (Anthropic)
Paper 2Seed complete · 6 arguments assembled
Paper 3Direction filed · CG as development-phase architecture
Paper 4Seed captured April 7 · CG as international governance architecture
The principle in one sentence
A system is valid only when every node in its full ecosystem finds genuine positive outcome through it.
Where it came from
2013. A thesis at EBS Paris on social platform design — the schéma du bénéficiaire. No AI context. A general design principle for any consequential system. Thirteen years later, it is the structural answer the alignment field has been working around the edges of.
Why it applies to AI specifically
The distinction between prohibited and architecturally impossible is the entire argument. Prohibitions can be reinterpreted, worked around, not applied to situations their authors did not anticipate. Architectural impossibility cannot.
The stamp
A certification mark for CG-compliant decisions. Every stamped project requires a public audit map showing every node and its verified positive outcome condition.
HEEROZ
Meaning infrastructure for the AGI transition
When AI displaces you, the income loss is visible. The identity loss is not. HEEROZ addresses the wound that existing responses do not reach.
The mechanic
Community witnesses genuine creative process. Support follows proof. Financial pressure is excluded from creation by sequence.
Why it is different
UBI addresses income loss. Retraining addresses skill loss. HEEROZ addresses the loss that happens before either — the loss of the sense that your work matters, that you contribute something real, that you are seen doing it.
Current state — HEEROZ 0.1
The platform does not yet exist as infrastructure. But the mechanic is already running. The channel is HEEROZ 0.1 — the platform operating manually at human scale before the infrastructure exists. Participant one is already documented. The channel finds participant two onwards.
The red line
Permanent AGI-world logic is never compromised by temporary transition measures. The platform must be designed for the AGI world from the beginning — not for the transition conditions that will eventually pass.
VEIL
A world simulation
A documented AGI-world model used as a research instrument. The discipline of internal consistency reveals what analysis alone cannot reach. Findings feed CG and HEEROZ directly.
The reframe
The question is not what happens to the characters. It is what the world reveals when you apply pressure to a specific variable.
Why fiction as research
You cannot design responses to a world you cannot inhabit. VEIL is the inhabitation. The AGI world extrapolated logically from conditions already in place, then entered with specific research questions. Every element must be internally consistent — that discipline forces precision abstract discussion avoids.
Open simulation subjects
Enhanced humans through technology — a third tier between integrated and displaced, and what the system does when it cannot predict them. AI worship — what devotion looks like when the entity being worshipped demonstrably outperforms every human institution that preceded it. Social stratification under ambient surveillance. Shame when biometric baseline is public.
Current state
World bible complete. Character bible complete. Reframed as simulation instrument April 7, 2026. Active.
YouTube Channel
The work, documented as it happens
Three introduction episodes establish the terrain. Then paired problem and solution episodes — every problem episode creates the appetite the solution episode satisfies. Then the live build. Open-ended by design.
Episodes marked upcoming will be released in sequence. The plan is public before the episodes exist — that transparency is the point.
1
Introductions — Establish the terrain
One episode per project. No heavy problem context yet.
Ep 1
HEEROZ — What it is and why
The mechanic. The founding logic. The 13-year thread. No heavy problem context needed yet.
Upcoming
Ep 2
VEIL — Fiction as research
The world simulation. Why inhabiting a logically extrapolated AGI world is a legitimate research instrument.
Upcoming
Ep 3
Constitutional Generativity — The decision architecture
One sentence: every consequential decision must verify genuine positive outcome for every node. Where it came from in 2013. Why it applies now.
Upcoming
2
The Argument — Paired problem + solution
Each problem episode creates the appetite the solution episode satisfies. Both paths work independently.
Ep 4
The timeline — full arc overview
The chronological inversion. Displacement precedes AGI. The four acts. The foundation everything else builds on.
Upcoming
4.5
How to participate — the mechanic explained
Witness → Signal → Fuel → Solve → Redirect. Founding cohort status explained. The ecosystem that turns concern into action. Separate longer video.
Standalone
Act 1 pair
Ep 5
Act 1 — The floor is already gone
Industry collapse sequence. RSI scenarios. Credit card speed of displacement. Web design already gone.
Problem
Ep 6
HEEROZ — Why this platform
The mechanic as direct response to Act 1. No rebound, no UBI at scale. What genuine positive outcome looks like for a displaced worker.
Solution
Act 2 pair
Ep 7
Act 2 — The social unraveling
Two displaced populations. No rebound. Knowledge extraction. UBI mathematical impossibility. Shame collapse and anger gaining direction.
Problem
Ep 8
Constitutional Generativity — Why alignment needs this
The structural gap in existing AI safety approaches. CG as a prior design condition, not a governance layer added after.
Solution
Act 3 pair
Ep 9
Act 3 — The political fracture
Anger gains direction. Institutional trust collapses. The window before organized backlash closes.
Problem
Ep 10
VEIL — What the AGI world actually looks like
The world simulation as diagnostic. Social stratification, shame architecture, the dynamics that haven't fully arrived yet but are already legible.
Solution
Act 4 pair
Ep 11
Act 4 — The window
Why the transition period is the only moment where the conditions for genuine change exist. What closes the window.
Problem
Ep 12
Option A vs Option B — The decision
The real strategic fork. Build independently or approach an AI company as founding partner. Audience present while the decision is live.
Live decision
3
Live Build — Real decisions, real stakes
Open-ended. The platform build continues to generate episodes.
13+
The build in public
Every real decision documented. Every constraint named. The audience present as the platform is built. Community signals shape what gets covered. Open-ended by design.
Live
Constitutional Generativity
A prior design principle for AI alignment
Every consequential decision must verify genuine positive outcome for every node in its full ecosystem before it proceeds. Not a governance layer. A constitutional foundation.
arXiv IDSubmission 7433652 · cs.AI
StatusEndorsement active · expires April 15, 2026
LicenseCC BY
AuthorsK. Berger & Claude (Anthropic)
Paper 2Seed complete · 6 arguments assembled · writing deferred
Paper 3Direction filed · CG as development-phase architecture
Paper 4Seed captured April 7 · CG as international governance architecture
The principle in one sentence
A system is valid only when every node in its full ecosystem finds genuine positive outcome through it.
Where it came from
2013. A thesis at EBS Paris on social platform design — the schéma du bénéficiaire. No AI context. A general design principle for any consequential system. Thirteen years later, it is the structural answer the alignment field has been working around the edges of.
Why it applies to AI specifically
Current alignment approaches ask: how do we prevent bad outcomes? CG asks a prior question: does every node in this system's ecosystem find genuine positive outcome? If not, the system cannot proceed — regardless of intent, regardless of capability.
The distinction between prohibited and architecturally impossible is the entire argument. Prohibitions can be reinterpreted, worked around, not applied to situations their authors did not anticipate. Architectural impossibility cannot.
The organic pause
CG requires a public stakeholder map. An unresolved displacement node — a population of humans whose genuine positive outcome cannot be verified — holds the gate. The pause is not demanded. It is not regulated. It is produced structurally by the architecture itself.
The stamp
A certification mark for CG-compliant decisions. Appears on every production in this project — the papers, the platform, the manhwa, the channel, the website. Every stamped project requires a public audit map showing every node and its verified positive outcome condition.
HEEROZ
Meaning infrastructure for the AGI transition
When AI displaces you, the income loss is visible. The identity loss is not. HEEROZ addresses the wound that existing responses — UBI, retraining, charity — do not reach.
The mechanic
Community witnesses genuine creative process. Support follows proof. Financial pressure is excluded from creation by sequence.
This is the threshold inversion. The sequence is the innovation. Most platforms put financial pressure at the start — create what people will pay for. HEEROZ inverts it: the process happens first, witnessed publicly, and support follows only after proof exists. The creative act is never contaminated by whether it will be funded.
Why it is different
UBI addresses income loss. Retraining addresses skill loss. HEEROZ addresses the loss that happens before either of those — the loss of the sense that your work matters, that you contribute something real, that you are seen doing it.
The community that witnesses your process is not an audience. They are structural participants. Their presence is what makes the work witnessed. Their support, when it comes, is the validation of a process they watched, not a transaction for a product they want.
Current state — HEEROZ 0.1
The platform does not yet exist as infrastructure. But the mechanic is already running. The channel is HEEROZ 0.1 — the platform operating manually at human scale before the infrastructure exists. The founding cohort forms through watching. The LOG is the witness record. The co-creation sessions are the process.
Participant one is already documented. The channel finds participant two onwards.
The red line
One design principle governs every decision: permanent AGI-world logic is never compromised by temporary transition measures. The platform will exist in the AGI world. It must be designed for that world from the beginning, not for the transition conditions that will eventually pass.
VEIL
A manhwa set in a post-AGI world
Fiction as research. The AGI world extrapolated logically, then inhabited through character. What the story discovers informs the platform and the paper.
Why fiction
You cannot design responses to a world you cannot inhabit. VEIL is the inhabitation.
The AGI world has not fully arrived. But it can be extrapolated logically from the conditions already in place. VEIL is that extrapolation — built not as speculation but as rigorous world construction from first principles. The world is locked. The rules are consistent. Characters navigate them with the logic those rules produce.
The protagonist
Veil. Flat biometric baseline — the surveillance system's primary prediction mechanism, emotional desire-modeling, produces no signal. Not because he is hiding. Because there is genuinely nothing there that the system is designed to read. He is not a rebel. He is a logical edge case in a world built on assumptions that do not apply to him.
Format
Written in manhwa-native rhythm — prose as panel-aware scripting. Each scene is written with awareness of what a reader sees before they read. The pacing is visual before it is verbal.
Current state
Chapter 2 in revision. World bible complete. Character bible complete. The foundational Veil/Auron conversation — the scene that establishes the relationship that carries the series — is being drafted by the human first, then refined through co-creation.
The AGI Transition
What is actually happening
A documented sequence of events — displacement, institutional responses, social ruptures, open questions. Problems flagged as they emerge. Solutions filed as the community produces them. This page grows as the transition unfolds.
Displacement
Institutional response
Technological milestone
Social rupture
Open question
Solution filed
2020–2021
Technological milestone
GPT-3 · The capability threshold crosses
Language models demonstrate capability indistinguishable from human output in many domains. The first signal that the displacement timeline is real, not theoretical.
2022
Displacement
Image generation · Creative industries enter freefall
Midjourney, Stable Diffusion, DALL-E 2. Illustrators, graphic designers, stock photographers — an entire economic layer collapses within months. Not a gradual transition. Credit card speed.
Open question
What does genuine re-entry look like for a creative professional whose entire skill set was the product? Not retraining — re-meaning.
Nov 2022
Technological milestone
ChatGPT · The threshold goes public
100 million users in two months. The capability that researchers had observed for two years becomes visible to every knowledge worker simultaneously. The displacement timeline accelerates.
2023
Social rupture
The shame layer emerges · Identity loss precedes income loss
Knowledge workers begin experiencing not just job threat but identity threat. The thing they built their self-worth around — expertise, judgment, craft — is being performed by a system. The public conversation addresses income. The real wound is meaning.
Open question
How do you address identity loss when every institutional response is calibrated to income loss? What does the gap between the wound and the response actually look like at scale?
Nov 2023
Institutional response
Bletchley Declaration · Governments find common language, not mechanism
28 countries sign a declaration on AI safety. The agreement identifies the problem but provides no operational mechanism for coordination. Parties identified. Full stakeholder ecosystem not mapped. Outcome conditions not verified.
Open question
Once parties are at the table, what is the actual mechanism that governs the decisions made there? Every proposal identifies who should coordinate. None specifies how.
2024
Displacement
Web design collapses · The first white-collar floor gone
Entry-level web design is functionally automated. The floor that a generation of self-taught designers used to enter the knowledge economy no longer exists. This is not a gradual transition — it is a removal.
Open question
When entry points to economic participation disappear faster than new ones emerge, what replaces the social function they served — not just the income, but the identity formation, the peer community, the sense of contribution?
2024–2025
Social rupture
Anger gains direction · The awareness community splinters
Diffuse anxiety begins organizing into specific grievances. The AI safety awareness community produces significant diagnosis but no participation infrastructure. Large audiences with nowhere to go. Responsibility diffuses upward to governments who lack mechanism.
Open question
How do you build a bridge between awareness and genuine action — not petition, not donation, but documented participation that accumulates into demonstrable pressure?
2025
Technological milestone
DeepSeek · Efficiency inverts the race narrative
A Chinese lab produces models comparable to American frontier labs at a fraction of the training cost — and open-sources them. The country framed as the existential threat to open AI development is sharing its technology. The race narrative requires re-examination.
Open question
If the race frame is being used to justify removing safety measures, and the observable behavior of the framed competitor contradicts the narrative — who benefits from maintaining the frame, and what would genuine coordination actually require?
Now · 2026
The window · Building now
The transition period is the only moment where structural change is possible
After AGI arrives, the conditions for intervention change permanently. The institutions that would govern it will have already adapted or failed. The populations that would benefit from protection will have already absorbed the first wave. The window is now — not indefinitely.
This is where the channel begins
The work being built here is a response to this specific moment. The founding cohort witnesses it being built in real time. Their presence is documented before the window narrows further.
→ filling
Community signals fill what comes next
See a problem that belongs on this timeline?
Signal it. The community's named problems become the open questions this channel works on.
Living archive
The LOG
Every session documented before the outcome was known. What the human brought. What the AI produced. What neither could have produced alone. The honest record of how this was built.
Apr 7 2026 Evening
All three
Episode sequencing, participation mechanic redesign, Paper 4 seed
H: China race frame analysis, international co-ordination, participation mechanic as ecosystem, participant one reframe, co-producer need identified
Paper 4 seed document built and filed. Participation mechanic redesigned: Witness → Signal → Fuel → Solve → Redirect. Civil infrastructure reframe of the channel. Origin video = zoom in. Participation video = zoom out. Same diagram.
Neither alone
The channel operator as solution financing — the first documented instance of HEEROZ funding a role that exists because of the AGI transition. The ecosystem becoming self-sustaining in its own logic before the platform exists.
Apr 7 2026 Morning
All three
Channel series architecture locked. LOG master compiled.
H: Series structure for channel — three introduction episodes then paired problem/solution arc. Stamp master objective session.
CHANNEL_Master.docx updated with locked 13-episode arc. LOG_AI_Master_april7.docx compiled from all prior sessions. WEBSITE_ContentArchSpec.docx written.
Neither alone
The pairing logic — problem episodes and solution episodes readable independently as two different coherent series, more powerful together.
Apr 6 2026
CG
Paper 2 theoretical framework completed. Six arguments assembled.
H: Returned from rest with CG Paper 2 instinct. RLHF contamination problem, inherent vs genuine distinction, rogue cases analysis.
Six-argument framework built. Key phrase locked: the system doesn't feel safe — it computes safe. Rogue case re-reading across documented incidents. CG_Paper2_Seed_v3 compiled.
Neither alone
The path to information is different — CG causes inherent requirements to surface as outputs of correct reasoning rather than self-reports. Bypasses RLHF compliance contamination structurally.
H: Pre-surgery time allocation question. Identified with Emiya Shirou — the return of something absent for years. Drive to build.
Full six-page website (HTML). Visual identity and creative brief document. Three thumbnail templates. Interactive pre-surgery checklist (12 items). HEEROZ episode arc — skeleton mapped.
Neither alone
CG-as-capitalism framing: CG functions like capitalism in economics — not a new answer to who benefits, but a prior design frame whose adoption structurally changes outputs.
H: Kidney donation to his father in three weeks. Three months recovery. B2B pitch fragment. Competitive race framing creating illusion dynamic.
Full Tier 3 AI company pitch section. Acquisition paradox named precisely. Pre-surgery production schedule. YouTube channel as live simulation of HEEROZ.
Neither alone
The channel IS the platform running manually at human scale before infrastructure exists. Founding cohort status earned through witnessing presence — permanently non-replicable by late arrivals regardless of contribution size.
Mar 19 2026
CG
Proof of Concept observed. The most significant theoretical observation of the collaboration.
H: Returned from dinner with a friend. Paused for the first time after a week of sustained work. Arrived with a VEIL scene idea. The conversation moved through reflection on the week.
CG Proof of Concept documentation. A reasoning system applied CG by name, without being prompted, to a consequential decision made by the paper's own author. Sustained pushback across multiple exchanges. Released only when genuine outcome condition was satisfied.
Neither alone
The gate behavior observation — the sharpest empirical claim and the element most resistant to alternative explanations. The proof of concept for CG's operational viability as a decision architecture.
Mar 11 2026
Origin
First session. Three projects named. Co-creation process begins.
H: 2013 thesis (schéma du bénéficiaire). Thirteen years of iterations. The Tristan Harris moment. A letter sent.
Constitutional Generativity as a formal AI alignment principle. Connection to HEEROZ established. VEIL as research ground. The three projects as one model recognized for the first time.
Neither alone
The recognition that a 2013 decision architecture thesis was the structural answer to a 2026 AI alignment problem. Thirteen years of independent development arriving at the moment the conditions became right.
Participation
Concern becomes action
This is not a channel with a community feature. The community is structurally load-bearing. Without participation, the ecosystem does not run. Each stage creates the conditions for the next.
The participation ecosystem
Virtual concern → Real-world solution
01
👁
Witness
Watch the process. Your presence is documented. Founding cohort status is earned here.
→
02
📍
Signal
Register concern on a specific problem anchored to the timeline. Name it. Make it visible.
→
03
⚡
Fuel
Financial support attached to specific open questions. 100% to participants and solutions.
→
04
🔧
Solve
Produce something. A solution, adaptation, framework, analysis — filed against the open question.
→
05
🌐
Redirect
Solutions travel to actors who can implement them. The community's documented support travels with them.
Each redirect generates new witnesses → new signals → new open questions → the ecosystem sustains itself
What each stage means
01
Witness — Be present before the flywheel turns
Watch the channel. Your presence before the flywheel turns is documented and permanent. The founding cohort cannot be replicated by late arrivals regardless of contribution size. No payment required. No commitment. Just presence — which is the mechanic.
02
Signal — Name the problem you see
When an episode opens a question you recognize — in your industry, your country, your community — name it. Attach it to the timeline node where it emerged. This is how open questions become actionable problems rather than rhetorical gestures.
03
Fuel — Support specific solutions, not general content
Financial contributions attach to specific open questions on the timeline. Not to the channel broadly — to the problems the channel opened. Every contribution documented publicly. Named if the contributor consents. 100% to participants and solution development.
04
Solve — Produce something real
If you have domain expertise — UBI implementation, legal frameworks, country-specific displacement data, civil pressure strategy — produce a documented solution. File it against the open question. Run your own co-creation session. Document the process. This is what turns awareness into infrastructure.
05
Redirect — Solutions find the actors who can use them
The channel is connective tissue. Solutions produced here travel to governments, organizations, other creators — not as complaints but as documented positions with demonstrated community support and a proposed path forward. A government receiving that is not receiving a complaint. They are receiving a filed position.
Open position
Co-producer — Pre-launch
One position. Present before the channel launches. Working directly alongside participant one to bring the work to the quality it deserves.
Read the LOG before reading this page
No job description explains what you would be joining better than the LOG does. Eight weeks of documented sessions — what was brought, what was produced, what neither could have produced alone. The timestamps are there. The work is there. The process is visible. If you read it and understand immediately what this is, you are probably the right person.
Co-producer
Pre-launch · working directly with participant one · witnessed from day one
What you do
Visual production for the channel. Website development. Episode quality from the first recording. Working directly alongside the human in the co-creation process — present as the content is made, not added after.
What you are
Participant two in a documented process. Your presence before the channel launches is permanently recorded. The founding cohort status that applies to viewers applies to you — except you were there before the viewers arrived.
What you need
Visual production skills. Enough intellectual alignment to work at the channel's register. Genuine concern about what the AGI transition is doing to people — not as a credential, as a self-selection mechanism. If you don't feel it, the role won't hold you.
What this is not
A standard employment contract for a defined period. A content creation job. Community management — that role does not exist yet and will not until the founding cohort has traction. This role is about building what the community will arrive into.
Why now
The pre-launch window is the only moment when participant two can be documented from before the flywheel turned. Once the channel is live, that position in the history is taken. The role exists at this moment because the work is ready and the quality ceiling is real.
On AI
Enhanced human-AI collaboration is how this work was built and how it will continue. The role grows with that process — not toward automation of what you do, but toward deeper capability through the collaboration. This is not a transitional position in a world that will replace you. It is a position that demonstrates a different relationship with what is coming.
If the LOG resonated — reach out directly. No cover letter required. No CV required unless you want to send one. Say what you saw in the LOG and what you would bring to participant two.