To me, "self-aware" is essentially being able to watch yourself think. I'm pretty sure GPT-3 can't do that, in any model.
[1] https://news.ycombinator.com/item?id=29576737
[2] https://toldby.ai/4kQNd-_tvUG
This was on the first davinci model
To me, "self-aware" is essentially being able to watch yourself think. I'm pretty sure GPT-3 can't do that, in any model.