"AI Ethics Officer." "Head of Responsible AI." "Trust & Safety Lead." These postings fill my feed now in a way they didn't two years ago.

It's tempting to ask: are we too late?

I think that's the wrong question. The better one is whether we even have a framework to hire these people into.

In 1942, Isaac Asimov began a series of short stories — later collected in I, Robot — built around three simple laws meant to keep artificial minds safe for the humans around them. He wasn't really writing about robots. He was doing something more useful: recognizing, decades before the technology existed, that capability without a governing framework is dangerous.

Here's the part most people miss. Asimov didn't write the Three Laws as a solution. He wrote them as a trap to demonstrate that even a well-intentioned framework can fail in practice. Nearly every story is a case study in how a clean, logical set of rules produces unexpected — sometimes catastrophic — behavior in the real world. His recurring character, Dr. Susan Calvin, isn't an engineer. She's a robopsychologist. Her job is to diagnose why the rules, faithfully followed, still went wrong.

That should sound familiar to anyone deploying AI in their business today. We are all becoming robopsychologists — trying to understand systems that follow instructions precisely and still surprise us.

So no, I don't think it's too late. But we keep asking "how do we control it?" when the harder question is "what should the rules even be?"

Over the next few posts I want to work through one piece of that — starting with something Asimov left out of his famous laws entirely.

(I've been thinking this through partly in dialogue with an AI system, which feels appropriately recursive. More on why that matters later.)