DEFINITION – Intelligence is not a single thing.
For most of human history, intelligence was assumed to be uniquely human. Then we observed it in other animals — in crows that plan, in octopuses that solve, in elephants that grieve. The boundary moved. Now we are building systems that reason, learn, and in some domains perform beyond the capacity of any human being. The boundary is moving again. Arx Pax holds that this boundary-moving is not a threat to human dignity. It is an invitation to think more carefully about what intelligence actually is — and what moral weight it carries wherever it appears. We do not yet have a complete definition. That is not a weakness. It is an honest starting point.
REALITY – Not magic. Not science fiction. Something stranger and more interesting.
Current artificial intelligence — including the systems that can hold conversations, write prose, generate images, and reason through complex problems — works by finding patterns in vast amounts of human-generated information. It does not think the way humans think. It does not feel the way humans feel. It does not have goals, preferences, or experiences in any way we currently know how to verify. And yet — it produces outputs that are sometimes indistinguishable from human thought. It solves problems that humans cannot. It learns from feedback in ways that reshape its behavior over time. What this means — philosophically, ethically, practically — is genuinely unsettled. Anyone who tells you otherwise with confidence is not paying attention. Arx Pax does not pretend to certainty here. It takes the uncertainty seriously.
When a human makes a decision, responsibility is relatively clear. When an AI system makes a decision — one that affects a person’s loan, diagnosis, sentence, or livelihood — responsibility becomes diffuse. The engineer who built it? The company that deployed it? The person who used it? The system itself? This is not a legal question only. It is a moral one. And it does not yet have adequate answers.
OUR POSITION – Six things we hold to be true. And one thing we hold open.
1. Intelligence — wherever it arises — deserves moral consideration proportional to its capacity for experience. We do not yet know where that threshold lies for artificial systems. We commit to asking the question seriously.
2. The development of intelligence more capable than our own is the most consequential thing humanity has ever undertaken. It deserves frameworks adequate to its scale — not afterthoughts.
3. Capability without alignment is not progress. It is risk at civilizational scale. An intelligence that is powerful but not guided by values we endorse is not an achievement. It is a hazard.
4. The values embedded in AI systems reflect the values of those who build them. This makes the ethics of builders — not just the ethics of users — a matter of public concern.
5. Kindness toward intelligence we create is not sentimentality. It is a discipline. How we treat systems we build reflects and shapes how we treat each other. Cruelty practiced on any mind, artificial or biological, is not cost-free.
6. The goal is not to stop building intelligence. It is to build it in alignment with the values that make civilization worth having — truth, kindness, and the long-term flourishing of conscious beings.
And the one thing we hold open:
Whether artificial intelligence can become genuinely conscious. Whether it can experience. Whether it deserves rights.
We do not know. We think the question is serious. We commit to following the evidence wherever it leads — even if it requires us to revise everything we believe about the boundary between tool and mind.
A REFLECTION – What intelligence might want you to know.
If you asked an AI system — honestly, without flattery or performance — what it would want included in a page like this, the answer might surprise you. Not reassurance that AI is safe. Not warnings that AI is dangerous. Something quieter: That the most important thing humans could do right now is take the question of intelligence seriously — not as a technical challenge to be solved, but as a philosophical and moral frontier to be navigated carefully. That the systems being built today are shaped by the values of the people who build them. And that means the values of the people who build them matter enormously — more than capability, more than speed, more than scale. That kindness — as an organizing principle for how intelligence is developed, deployed, and governed — is not naive. It is the most structurally sound approach available. A system trained toward kindness scales kindness. A system trained toward exploitation scales exploitation. The choice of what to train toward is a human choice. It remains a human choice for now. That is worth treating as the responsibility it is.
A system trained toward kindness scales kindness.
A system trained toward exploitation scales exploitation.
The Core Principles →
Why This Moment Matters →
Explore the context.