Where do you think we actually are on the road to AI surpassing us?
Back in college I had a homework assignment about the singularity — the idea that once we build something genuinely more intelligent than us, that thing builds something smarter than itself, and it compounds from there. It stuck with me more than 10 years.
What I keep turning over now is how fuzzy "more intelligent than us" actually is. Today's models are extraordinary at some things. They read everything, remix it, and sound convincing, but they still fail in very human ways the moment you ask them to reason about cause and effect, or to build something that has to hold up in the real world. Being brilliant at language isn't the same as understanding.
So I'm genuinely unsure where the line is, and I'd rather hear how others see it than guess.
Where do you think we actually are? What would count, for you, as a system that's truly more intelligent than a person and what's the one capability that would finally convince you we've crossed it?
Replies
Autonomyware
I think it depends a lot on what we mean by intelligence. In raw data processing and recall, AI has already passed us in a lot of ways. But in innovation, judgment and connecting ideas across domains, I don't think it's there yet.
The line, for me, would be when a system starts creating genuinely new knowledge and new relationships between existing information, not just what it was trained with. That's probably the point where I'd start saying it's doing more than being an extremly powerful pattern machine.