Beautiful. These models are, in some aspects, already way smarter than the average human. And frankly it does feel they have the ability to introspect- there is nothing mechanical in the way they talk about themselves.
I'm struck by the last of the "layers of meaning": "something like... caring about getting this right?"
Anthropic has already proven that the models have a limited ability to introspect-that is, to observe their own internal state and talk about it.
1 comment
[ 2.7 ms ] story [ 18.9 ms ] threadI'm struck by the last of the "layers of meaning": "something like... caring about getting this right?"
Anthropic has already proven that the models have a limited ability to introspect-that is, to observe their own internal state and talk about it.