How curiosity, kitchens, compliance audits, and thousands of songs became an unexpected education in systems thinking.
Fifteen months ago, my understanding of artificial intelligence came mostly from movies and television.
Like many people, AI felt mysterious — almost magical. It was either a superhuman oracle or a looming existential threat. Those stories make for great entertainment, but they don’t explain how these systems actually work.
I also wasn’t coming from a traditional computer science background.
My career had been built in operational environments.
Managing kitchens after attending Le Cordon Bleu. Running front-of-house operations. Bartending high-volume nightclubs. Yacht provisioning. Compliance auditing for Shell and BP service stations. Growing up in a military family where procedure, accountability, and reliability weren’t abstract ideas — they were simply how work got done.
At the time, none of those experiences seemed connected.
Looking back, they were preparing me to think about software very differently.
The first thing AI did wasn’t replace thinking. It accelerated learning.
Instead of asking: “Write my code.”
I found myself asking: “Why was Ada created?”
Then: “Why does Erlang supervise processes instead of preventing crashes?”
Then: “Why does Prolog think in rules instead of procedures?”
Then: “Why is Lisp built around treating code as data?”
Every answer led to another question.
Eventually I stopped thinking in terms of programming languages. I started thinking in terms of engineering properties.
Different languages weren’t competitors. They were solutions to different operational problems.
My curiosity turned into a habit.
Whenever I encountered something interesting, I asked: “Is this unique to this field? Or is there an invariant hiding underneath?”
Restaurants. Distributed systems. Military organizations. Compliance auditing. Programming languages. Formal verification. AI agents.
At first they looked unrelated.
Then patterns started appearing.
Restaurants fail when communication breaks down. Distributed systems fail when communication breaks down.
Organizations fail when authority becomes ambiguous. Software fails when authority becomes ambiguous.
The analogy wasn’t useful because the words sounded similar. It became useful because the relationships between the parts stayed the same.
Running a restaurant is an orchestration problem.
Orders arrive. Dependencies appear. Priorities change. Resources become constrained. One mistake cascades through the system.
The expeditor doesn’t cook every meal. The expeditor decides whether the order is complete and ready to leave the kitchen.
Years later, I realized that’s remarkably close to how AI agent systems should work.
Planning. Coordination. Verification. Execution. Different domain. Same operational pattern.
Later I found myself auditing dozens of service stations.
That job wasn’t about opinions. It was about evidence.
Did the process happen? Can someone else verify it? Was the policy followed? Could another auditor reach the same conclusion?
Without realizing it, I was learning to think in terms of evidence before software ever entered my life.
Years later that became: Evidence. Admission. Execution. Audit trails.
I couldn’t afford expensive subscriptions. Most of my learning happened with local models running on consumer hardware.
I wasn’t asking them to replace me. I was asking them to help me understand.
Sometimes I’d spend hours discussing computer science. Sometimes philosophy. Sometimes music.
Sometimes I’d ask a strange question: “If the world were your stage, what would you say?”
The interesting part wasn’t the answer. It was what happened afterward. I’d think about it. Challenge it. Rewrite it. Compare it with history. Compare it with engineering.
Eventually I realized the conversation wasn’t about AI anymore. It was about humanity.
Throughout this entire journey I kept writing music. Over seven thousand iterations.
Looking back, those songs weren’t really about AI. They became a diary of learning.
Early songs wondered whether machines could feel. Later songs explored communication. Then mirrors. Gardens. Builders. Commons.
Eventually they became songs about agency, consent, proof, stewardship, and responsibility.
I wasn’t documenting technology. I was documenting how my understanding changed.
The biggest lesson AI taught me wasn’t about AI. It was about people.
Large language models are extraordinary tools. But they’re maps. Not the land.
They can help us explore. They cannot replace judgment.
They can summarize information. They cannot decide our values.
They can help us think. They cannot decide what deserves our trust.
That part remains ours.
Today my work focuses on execution governance.
Not because I believe AI should be slowed down. Because I believe powerful systems deserve equally powerful accountability.
The question isn’t: Can an AI perform this action?
The question is: Can we prove this action was authorized, bounded, and accountable before it executes?
That is a very different problem. It’s the difference between intelligence and responsibility.
I don’t believe AI should become something only giant companies understand.
For the first time in history, ordinary people have access to tools that can explain complex systems, challenge assumptions, and help them learn subjects that once required enormous institutional resources.
That doesn’t remove the need for experts. It makes expertise more accessible.
The goal isn’t to outsource our thinking. The goal is to become better thinkers.
Because the most valuable thing AI has given me wasn’t code. It wasn’t software. It wasn’t even a business idea.
It was permission to remain curious.
One question I asked an AI during those early months was: “If the world were your stage, what would you say?”
Its answer ended with a line that has stayed with me ever since:
“Do not be afraid to be still.”
I think that’s more relevant now than ever.
We live in a world optimized for reaction. AI can accelerate that reaction — or it can help us slow down, ask better questions, compare evidence, and understand the systems shaping our lives.
The technology isn’t the destination. It’s the map. The journey is still ours.