Everyone writes prompts; few write good ones. The secret isn't longer prompts, it's cleaner ones—prompt like a caveman. This talk works through two live examples that prove the point from opposite directions:
Image Classification: we'll build a working image classifier from scratch using nothing but prompts and show how the same task produces wildly different results depending on how precisely you state your intent.
Code Explanation: we'll point a model at an unfamiliar codebase and ask it to explain itself. Well-designed systems that are legible to humans are legible to models too.
Whether…

Hosted by Itai Katz, Will Angel & Avithal Elias Lautman