Not sure how much I am saying the same idea using different words, but it seems to me that LLMs are like System 1, and the things we put on top of them (skills, harnesses) are attempts to build System 2.
This may sound wrong, because in humans, we wouldn’t call e.g. programming a System 1 skill, but I think that’s just an example how the LLMs are already superior to humans in some aspects.
Not sure how much I am saying the same idea using different words, but it seems to me that LLMs are like System 1, and the things we put on top of them (skills, harnesses) are attempts to build System 2.
This may sound wrong, because in humans, we wouldn’t call e.g. programming a System 1 skill, but I think that’s just an example how the LLMs are already superior to humans in some aspects.