This is a super wrong take. The machines did not teach themselves. We fed them nearly the entirety of human-accumulatef knowledge and developed a math to allow the traversal of connections between those data points
The "unknowns" you're referring to are just properties of scale not fully explicable to us. Feed some LLMs enough languages, they can infer others. We know most of why this is - languages have far more in common than they dont - we just can't explain it to scientific certainty
Replies
No replies yet