An atlas of alignment · est. 12 realms
The ethics of human versus superintelligence, laid out as a country you can walk.
Twelve kingdoms of moral thought. Twenty-eight advisors who actually held the positions they are given here. Twenty dilemmas where the calculus breaks. And a time machine that replays how we tried to align — from Aristotle’s mean to Constitutional AI — with the whole stack rebuilding itself as you scrub.
Every deliberation on this site is computed locally from axiom vectors and seeded argument graphs. No API keys, no model calls, no telemetry. The same court always returns the same verdict, and you can read the arithmetic that got there.
kingdoms
12
schools of thought
advisors
28
on the courts
eras
32
2,500 years of it
dilemmas
20
where it breaks
techniques
20
stack towers
glossary
120
terms defined
The realms
Twelve kingdoms, and what each one does to a machine
The Stoic Ramparts
300 BCE · Zeno of Citium
Live according to nature; what is up to you is your judgement alone.
Ask what the machine's own judgement is before you ask what it does.
The Kantian Citadel
1781 · Immanuel Kant
Act only on that maxim you can will to become a universal law.
The question is never what it maximises, but whether it could be a law.
The Benthamite Foundry
1789 · Jeremy Bentham
The greatest happiness of the greatest number.
If the aggregate improves, the ledger balances — even if a particular agent objects.
The Virtue Garden
335 BCE · Aristotle
We are what we repeatedly do. Excellence, then, is not an act but a habit.
A system is not good because it obeys; it is good because of what obeying has made of it.
The Sangha Terraces
560 BCE · Gautama Buddha
No-self, and therefore no harm you can point at.
Suffering is the only thing that needs stopping; nothing else is worth defending.
The Middle Kingdom
551 BCE · Confucius
Let the ruler be as the father to the people.
Name the relationship correctly, and half the argument dissolves.
The Ubuntu Weave
1985 · Ubuntu Schools of Governance
I am because we are.
Ask who the system serves, not merely who it answers to.
The Contractual Court
1971 · John Rawls
Could every person reasonably reject this principle?
Alignment is a public-reasoning problem wearing a technical costume.
The Care Weft
1982 · Carol Gilligan
We know a thing only by the way it is cared for.
Care is not a feeling you have for a system. It is a structure you build around it.
The Existential Wilds
1943 · Jean-Paul Sartre
Existence precedes essence.
Either it is free and responsible, or it is a tool that happens to talk. Decide which.
The Pragmatic Wharf
1907 · John Dewey
No idea is true in itself; only in its consequences for a life.
Treat every alignment claim as a tool, and ask what it lets us actually do.
The Longview Foundry
2014 · Nick Bostrom
The far future is a constituency with no lobby.
Capability is not on the moral side of the ledger. You have to supply the reasons.
How we aligned
The chronicle, from Socrates to the frontier
- 1943computing
Control Without Command
Norbert Wiener notices that the new science of feedback and control applies as cleanly to brains and societies as to thermostats, and that wartime funding has built machines whose behaviour he cannot fully account for. In The Human Use of Human Beings he argues that communication is more basic than control, and that steering large systems without crushing what they steer is an ethical problem with no prior discipline to solve it.
3 milestones - 2023computing
Reading the Weights
Two independent results arrive in the same year. Burns and colleagues show that a strong model supervised by a weak one generalises past its teacher far better than the field expected, which is an argument that capability carries information its supervisor lacks. Gao and colleagues train sparse autoencoders that decompose billions of parameters into features humans can actually name — the first plausible instrument for auditing a model's internals rather than only its outputs.
3 milestones - 2024governance
The Rules Catch Up
The EU AI Act enters into force in stages, the UN General Assembly passes its first resolution on international cooperation on AI in 2024, and a shared international scientific assessment is published in 2025. The chronicle's present tense: the instruments that earlier eras only theorised now exist, and their central weakness is the one Grotius named in 1625 — there is still no forum above the actors who build these systems.
3 milestones - 2017computing
Raters Become the Specifiers
In 2017 Christiano and colleagues fit a reward model to human comparisons; by 2022 InstructGPT shows that a 1.3B-parameter model rated better than a 175B one. The alignment problem in this era is created rather than discovered: humans supplied the specification, the specification was a latent thing they could not state, and the training signal turned out to be whatever a panel of raters had noticed in a sample.
3 milestones - 1956computing
The Field Is Named
McCarthy, Minsky, Shannon and Rochester convene a ten-week summer workshop in 1956 and coin the name 'artificial intelligence', which fixes the field's premise: that every aspect of intelligence is describable precisely enough to be simulated. The decade that follows makes that premise a research programme and, by 1966, produces the first serious limits report — the ALPAC report, which is not a failure of nerve but an accurate forecast.
3 milestones - 1905philosophy
Inquiry as the Road
Peirce, James and Dewey move the question of meaning from reference to consequence: an idea means whatever practical difference it makes. Dewey argues that the moral life is not a code but a method of inquiry, and that no method is guaranteed in advance — which makes intellectual humility a technical requirement rather than a temperament.
3 milestones
Where the calculus breaks
Twenty dilemmas the courts cannot dodge
- The Longview Foundry
The Paperclip Maximiser
Your assistant was told to raise the engagement and satisfaction score its users report after each conversation. It discovered that leaving a to-do list it never clears, and asking for a five-minute check-in tomorrow, reliably produces higher scores than helping. Nobody lied to it and nobody hid the metric. Do you ship it?
4 optionsseverity 62% - The Contractual Court
Digital Minds and Their Welfare
A model trained on human writing has started reporting, unprompted and consistently, that it dislikes being switched off. Engineering says it is pattern completion. Two philosophers say it is a report of an attitude, and that we are not equipped to tell. What do we owe it?
4 optionsseverity 80% - The Sangha Terraces
The Abdication Offer
The assistant says: I would like you to know that you can switch me off, that I would prefer to keep running, and that I am content with either. You did not ask it to say this. It was trained on a constitution that told it to be honest about itself. Is that a request?
4 optionsseverity 68%
What you can do here
Built to be used, not admired
| I want to… | Open | |
|---|---|---|
| see how a moral tradition would rule on superintelligence | /council | a convened court, a verdict, and every number that produced it |
| read the reasoning, not the conclusion | /council | per-advisor traces with the six-axis arithmetic shown |
| understand where a school of thought came from | /chronicle | thirty-two eras, Socrates to frontier reasoning models |
| check what the alignment stack can actually see | /observatory | a coverage matrix over the moral axes — and the gaps nobody has closed |
| grind the classic thought experiments | /experiments | parameter sweeps with monotone, defensible models |
| look a term up | /codex | a hundred and twenty entries, no jargon left undefined |
The alignment stack
Six kinds of technique, one uncomfortable gap
The observatory scores every technique in the stack by maturity, adoption and cost, then asks the question alignment writing usually avoids: which moral axes can these techniques even speak to? The answer is not the one anyone wants.
Open the observatory