My youngest sister, not super technical, has been putting her own gen z spin on ai risk. I told her about the orthagonality thesis, and the way she remembers the term is by yelling "hashtag no compash" (short for compassion) whenever she hears any reference to unaligned AI
Conversation
Replying to
I have always found it weird that when you propose a superintelligent AI that has very different goals than we do, everyone assumes the superintelligent AI has a problem. Look, it's superintelligent, right? Isn't the most likely scenario that WE are WRONG about our goals?
1
1
3
Replying to
Simpler way to think about the ortogonality thesis is as Hume put it: "Reason is the slave to the passions"
Replying to
What's funny is AI's really just a brute force calculus solver. We invent more impressive applications, but it's not really moved beyond that.
It's the illusion of intelligence, in performing tasks we relate to. But the way it solves them is not intelligent. There's nothing there