Today I had a little epiphany. I was writing some code with ChatGPT and I realized that doing that is like being a manager and having someone reporting to you.

Today I had a little epiphany. I was writing some code with ChatGPT and I realized that doing that is like being a manager and having someone reporting to you. You explain your assistant what they need to do and they come back with the code a while later. Except that this assistant is way dumber than a person. But also way, way faster. The assistant makes mistakes, lots of mistakes. But you point out the mistakes, and either they fix them, or eventually you fix them. In the end, if you are a good manager, you get things done. And you get things done 10 times faster than with a human assistant. Most companies did not figure this out yet. They are concerned that there could be security or privacy vulnerabilities. Or maybe they simply didn't play enough with AI to figure out the best way to extract value. Maybe they think AI is some magic pixie dust that will write a Richard Rhodes-level of a non-fiction book. Or they hope that it will evolve to become something like that. It may or it may not, but right now AI is far from this. But some companies will figure this out sooner or later.