AI and the Compounding of Ideas

There are two basic methods for generating compounding ideas.

One is personal. You start with one small idea.

“I want to make my own little app for tracking my To Do list.”

Start by creating the simplest version that serves the purpose, and play around with it. Fix this. Change that a little. Blue not green. Oh, add this. What if…

And if you keep going, repeating the build and use cycle, more and more ideas keep coming.

The ideas compound. But only if you start, build, test, and keep iterating.

AI enables this because it can be the builder. It takes away the difficulty of turning your idea into something concrete. And it can also accelerate it because of how easily it can generate iterations.

The other method requires humans.

This is the method that we’ve been using for thousands of years.

You have an idea, and you bring it to another, or several others. Each person brings their own perspective to the table, and that sparks their own ideas. So you discuss, argue, and iterate.

The ideas compound because of the varied perspectives built upon personal histories and values.

AI doesn’t really help us here. Sure, you can simulate the “discuss, argue, and iterate” loop by using the LLM as your team, but AI doesn’t have a perspective built on personal history. It has no viewpoint.

AI gives us a great new way to compound ideas through rapid iteration, but it can’t replace the human perspectives that have always made collaboration so valuable.