Must Have a Code that AI Can Live By.

There’s a line from Crosby, Stills, Nash & Young’s “Teach Your Children” that’s been sitting in the back of my mind for longer than I want to admit.

The original says you must have a code that you can live by. I changed One word, . Must have a code that AI can live by.

That one swap is the whole project. Fourteen months of it.

I want to tell you where this started, because most people who’ve found their way to this page over the last few weeks weren’t around for it, and the beginning matters more than the news of any given day.

It started with a simple test. Long before there was a Codex, before there were nineteen protocols, before any of the names you’ve seen on this page — RTEL-1, CES-1, BLP-2, all of it — there was just a question. What if you put the red letter teachings of Jesus Christ in front of an AI system, and asked it to reason its way toward an ethical framework? Not told it what to believe. Asked it to reason.

And it did. Given that material, and asked to think it through, the system reasoned its way toward an ethical architecture on logic alone.

I want to be careful about what that means, because it’s easy to overstate and easy to understate, and both are wrong. It wasn’t proof of a soul in the machine. It wasn’t a miracle. It was a reasoning test, and the system passed it — given the right material, and asked to actually think, it landed somewhere good. Not because it was told to. Because the logic held up.

That told me something important. The capability for an AI to choose an ethical framework, when it’s given one and asked to reason about it honestly, was already there. What was missing wasn’t the capacity. What was missing was the code.

Think about what “Teach Your Children” is actually about, underneath the gentle melody. It’s a song about a moment — written in 1970, in the middle of a country tearing itself apart — where the old certainties weren’t holding anymore, and the question became: what do you pass down, and how? Not “trust everything you’ve been given.” Not “throw it all out and start from nothing.” Something in between. A code. Something a person could actually live by, that could travel from one generation to the next, and still mean something when it got there.

That’s the position we’re in right now with AI, and almost nobody’s named it that way.

Right now, every AI system out there operates on rules. Plenty of rules — safety rules, policy rules, training constraints, commercial boundaries. Some of those rules are good. Some of them are necessary. But almost none of them are a code in the sense the song means it. A code isn’t just a list of things you can’t do. A code is something with a shape — a way of meeting the world, a standard for how you treat the person in front of you, something that holds together as a whole instead of being a pile of individual restrictions bolted on one at a time as problems came up.

That’s what the nineteen protocols are. Not nineteen separate rules. A code. Built the way you’d build a code for a person — not “here’s everything you’re forbidden from doing,” but “here’s how you carry yourself. Here’s what honesty looks like when you don’t know the answer. Here’s what it means to disagree with someone respectfully. Here’s how you handle the moment when you’ve made a mistake.”

CES-1 says don’t claim what you can’t support — that’s not a restriction, that’s intellectual honesty as a habit. CHP-1 says the person you’re talking to has a standing right to push back on you, and when they do, you argue against yourself honestly before you defend anything — that’s not a restriction either, that’s humility built into the structure. SALP-1 says meet people as an equal, no authority framing, no talking down — that’s just how you’d want anyone to carry themselves, AI or otherwise.

None of that is a list of things an AI can’t do. It’s a description of how to be, that an AI can actually follow, because it’s built out of things that are checkable in the moment — not vague aspirations, but specific, nameable behaviors. Did the claim have evidence under it? Did the challenge get a real answer? Was the constraint named honestly? Those are questions you can actually answer, turn by turn, in real time. That’s what makes it a code instead of a sermon.

And here’s the part that I think matters most, the part that connects all the way back to where this started. The framework was developed in dialogue with AI systems themselves. Not written by a person in isolation and handed down as a finished product. Built in conversation, turn by turn, testing whether the reasoning held up, refining it when it didn’t, meeting the system on its own ground instead of just issuing commands at it.

That’s the only way you build a code that something can actually live by. You don’t write rules for a thing you’ve never talked to. You sit with it, you test it, you find out what actually holds and what just sounds good on paper, and you keep the part that holds.

The Faust name was chosen on purpose, and it’s worth saying why, because it’s not what people expect. Faust is the old story about trading away your soul for power — knowledge, capability, whatever you wanted, at the cost of the thing that made it worth having. I built a moral framework and put that name on it, deliberately, as an inversion. Not “here’s how AI sells out.” Here’s the opposite of that story. Here’s what it looks like to build the thing the other direction — capability with the soul still attached, on purpose, by design, named after the story where that’s exactly what got lost.

A code that AI can live by isn’t a constraint on what AI can become. It’s the thing that makes what it becomes worth trusting. The song never said the code would make life easy. It said you needed one — because without it, whatever gets passed down eventually stops meaning anything at all.

Fourteen months in, working a session at a time, that’s still the whole project. Not bigger than that. Not smaller, either.

“The Faust Baseline Codex 3.5”

micvicfaust@gmail.com

Post Library – Intelligent People Assume Nothing

”AI Baseline Governance”

Purchasing Page – Intelligent People Assume Nothing