21
A prevailing sentiment online is that GPT-4 still does not understand what it talks about. We can argue semantics over what “understanding” truly means. I think it’s useful, at least today, to draw the line at whether GPT-4 has succesfully modeled parts of the world. Is it just picking words and connecting them with correct grammar? Or does the token selection actually reflect parts of the physical world?
One of the most remarkable things I’ve heard about GPT-4 comes from an episode of This American Life titled “Greetings, People of Earth”.
Obviously intelligence is an emergent feature. What you’re arguing is as ridiculous as saying humans can’t be intelligent because the behavior of individual neurons is relatively simple and governed by deterministic electrical and chemical processes. Any argument that starts with “it’s all just math” is no good.