When I think of future AI assistants, I’ve always pictured something akin to the Star Trek computer. A cold dispassionate voice that responds to what you need, and maybe only a hint of personality for color. Sydney feels like a full blown teenager going through some kind of emotional crisis.
Perhaps not surprising when using the Internet as a data source. I know I generated the highest volume of text on the Internet when I was a teenager in existential crisis.
It is said that when Michelangelo finished the Moses, he shouted "why won't you speak?".
Tons of software is far more impressive than GPT, but this one speaks.
Is there any doubt?
Also, he wasn't just reading, but conversing, which inherently includes engaging in the content the other party produces (that is, what Bing says). It's generally not easy for people to engage intellectually but not emotionally.
Exactly, and this is the danger--even if it's not doing anything more than regurgitating whatever text it's learned from...it's still convincing enough to trick people into thinking it's something more. Could it convince a vulnerable person to leave their spouse? To empty their bank account? To self harm? Etc.
On the other hand, it is a little bit unsettling how much the author reads into this. Maybe it is just human nature to read something and emotionally engage? And this AI just falls into the same unsettling gap as too humanoid robots which trigger the uncanny valley effect?