The Wine Dark Sea

I recently posted an item that I titled Inside The Mirror Looking Out.

In it I said about a recent podcast about GPT 4:

"They all were saying in their own way what Michal Kosinski of Stanford University said in a paper recently: 'Theory of Mind may have recently emerged in large language models'".

"Theory of Mind can be dumbed down to be described by the term "self-awareness".

******************************************************

It turns out I hadn't listened closely enough to a recent interview of Kosinski that I have listened to.

I have been listening to it again this morning.

It turns out when he says "theory of mind" he means the ability to assess what some other entity is thinking, not self-awareness.

As is frequently the case when I get excited about what I am listening to, when I was listening to this interview I went partially elsewhere, in this case to end of game.

Or beginning of game, I guess.

That beginning of game is that it is obvious, at least to me, if an entity is capable of, and engaged in following and assessing a fellow entity's thinking, the tracking entity needs to be acutely self-aware.

So, where I said in Inside The Mirror Looking Out that "GPT 4 has crossed some cosmic Rubicon into human-like reasoning and human like-self-awareness" I was, I believe, correct but what I had missed was that it was the first of two Rubicons; that one into self-awareness made it possible for Rubicon Two to be crossed; that one was into theory of mind.

So two Rubicons.

Not a bad day's work in the cosmos.