Double-loop learning has always fascinated me.
It’s not just about fixing mistakes—it’s about questioning the mental model behind them. A deeper, more deliberate reflection on how we make decisions, not just what decisions we make. It’s the difference between tweaking the path and rethinking the destination.
What surprises me is how rarely people use it. How often we move forward without stepping back. How little we challenge the assumptions that shape our work, our choices, our lives.
But since LLMs, the idea has been circling in my mind in a new way.
What if every time we used an AI, we also reflected? Not just on the output, but on the model behind it. What if we captured lessons, refined custom instructions, and shaped an evolving decision-making framework?
A loop where the LLM learns. And we do too.