What does it mean when people say that not even the creators of large neural networks or AIs like ChatGPT or Midjourney fully understand how they work?

Question

I’ve heard this sentiment a fair bit in recent years and I’ve never been able to find a good answer online.

In: Technology

Answer 1

It’s not that they don’t understand how they work, only that they don’t know every line of code like a traditional program.

AI like this learns. It’s fed millions of documents or images and told what each is or something like that. Eventually, it can recognize what an apple looks like etc. This only happens because it can change it’s own programming. Because of this, it’s impossible for anyone to be able to predict exactly what it will do, or exactly what steps it will take to accomplish a task.

The original code was written by humans, but what the AI becomes has parts that have never been touched by a person.