> I want to understand how things work. AI feels like a black box to me. The amount of papers I’d have to read and mathematics that I’d have to ingest to really

> I want to understand how things work. AI feels like a black box to me. The amount of papers I’d have to read and mathematics that I’d have to ingest to really understand why a certain prompt X results in a certain output Y feels overwhelming. How is this a legitimate argument? We have invented so many algorithms and used so much math to build our software along the years. They are not easy to understand. Understanding them requires tons of effort, including reading papers. People may have forgotten that 30 years ago, the "host stuff" was still systems, and people did read "deep" papers. People still do nowadays, except that only the very experts do so. Besides, the math is really not that hard -- merely college level. In contrast, go read an introductory book on program analysis or type system or distributed algorithms. Those maths can be harder as they are more abstract. In addition, the amount of code in a model is orders of magnitude than a compiler or a distributed system or game engine and etc. I'd argue that it's actually easier to understand how a model works.