The distinct learnings
The LLMs demonstrate a great intelligence regarding responses. Is that really intelligent? Is intelligence obtained like that?
Chip discusses in the book about that classic frame, that some tasks are simple for humans and hard for machines and some tasks are hard for humans and simple for machines.
To go beyond to understand the universe, is it necessary to learn like a human?
For sure, some traits in the human intelligence are really necessary for a better reasoning. But, in terms of reasoning, specially, the reasoning models are there, and papers demonstrating what looks like a reasoning step.
Is it just scaling?
Interesting. Working with Claude. Living with my son. Observing intelligences. It is a mystery for the machine and human. Neural networks are criticized because of the lack of understanding. But I remember that professor questioned this in a discussion, that we also do not understand the human learning, and if this is the question. Really interesting to consider that. Great learning.