AI is the ultimate rubber duck

In software development there’s a concept called rubber duck debugging. You explain your code from start to finish to an inanimate object, classically a rubber duck. In doing this you’ll often figure out some problem in your code.

Rubber duck debugging isn’t only useful for programming.

AI is amazing. It’s great for research and exploration. It can help you understand new subjects and find that movie you can’t remember the name of.

I wouldn’t blindly trust AI to give me life or business advice. AI is a huge yes-man. It tends to give boilerplate output. But I still run these questions by it.

The act of writing to AI is where I get a lot of the value.

Explaining a thought or concern from start to finish is useful for all kinds of situations. We jumble all these thoughts together and become blinded by them. Forcing yourself to step back is why this works.

Something about the conversational form of the writing helps break it apart. You need to start from a clean slate and explain it in a way that another person could understand it. As you describe the problem you understand it more clearly.

I’ll often have epiphanies before I ever submit the prompt.

The weird thing about this is that AI is most useful when I need it the least. The answer is already in my head but I need to lay out my thoughts to see it.

Rubber ducks don’t solve anything here. Neither does AI. I just need something to talk to and AI is always there. The yes-man doesn’t really matter because I wasn’t really asking it anyway.