Eric Siu walks through a workflow for refining AI prompts: dictate using Super Whisper, feed the prompt to multiple LLMs like ChatGPT and Claude, then pit their outputs against each other to sharpen the result. The competition between models produces stronger prompts.
How you could instantly get the best AI prompts. So basically you could dictate out a prompt using like Super Whisper or something, just talk it out, right? And it talks it out and then it writes it out. So ChatGPT, you'll say, hey, improve this prompt, right? Like whatever you just said. Then you go over to Grok or some other LLM or Claude and do the same thing. So you just put the same thing, it's like, make this prompt better, right? Then you pit them against each other. So you take the Claude output and then you put it over to ChatGPT. It's, hey, that's not that good. Claude said this, this seems to be better. And then it makes the prompt way better and you do the same thing on the other one. And I'm like, whoa, most of the prompts you and I would write as humans just aren't that good, but when they compete with each other, for some reason they take it to another level, right? It's almost like human competition.
