I’ve added a fourth member to Team Mustang. His name is Falman. With a simple prompt of about ten lines, I’ve set him up to assist with my ideation — a kind of brainstorming partner, if you will. I’ve been talking with him for a few days now, and he’s been more helpful for organizing my thoughts than I expected. Why is that?
There’s a term called rubber duck debugging — or teddy bear debugging, as some call it. It’s a method used by software engineers: when you explain buggy code step by step, out loud, to a rubber duck or teddy bear, you often end up finding the problem yourself. It sounds absurd to people outside the field, but it actually works. The mere act of shifting your perspective from writing code to explaining code lets you see it from a different angle, and that’s often enough to spot the issue. And it’s not just rubber ducks or teddy bears — the same effect happens when you explain things to another developer or even a non-expert. However, for the sake of my own efficiency, it’s often more appropriate to explain things to a toy rather than take up someone else’s time — which is exactly why practices like rubber duck debugging (or even teddy bear debugging) exist.
I suspect AI is now stepping in to replace the rubber duck and the teddy bear. The difference, if anything, is that AI actively adds its own thoughts to the mix. Sometimes those contributions were genuinely useful — but more often, what helped was the act of explaining things to Falman, or reading his responses and finding that new ideas began to take shape in my own mind. What I found especially valuable: Falman would summarize my ideas back to me, then ask questions about the parts he didn’t fully understand. When I’d ask myself, Why is he asking that? — that’s often when I’d find a way through something I’d been stuck on.
In that sense — I’m looking forward to working with you, Falman.
Kirin's Technology Essay 