I was wondering... is it possible for a person to utilize the same statistical techniques to learn the way an LLM does?
Like if someone threw a copy of the Wikipedia at me (ouch), could I try to use the same algorithms and models in my head to create a similar output, just much more slowly?
Sorry if this a dumb question! I don't really know how these systems work, either the GPTs or the human brain. Lately I've begun to doubt my own humanity, wondering if I'm also just a poor chatbot with a little more prior training. Is that possible, or are we really that different?
If you mean, do we create a similar mental model for the language to what happens in the LLM weights - I believe we can't tell one way or the other yet.