The article never substantially mentions the subject declared in the title, and conflates it with Intelligence - whihc is something completely different.
> [AI] was conceived, if memory serves, at Carnegie Mellon University in the 1970s
I am still for Dartmouth '56. What is the author pointing to?
The author seems to be extremely abstract about the stated point, and seems to be instead interested about building up a narrative about economic control.
I would say that LLM-based AI will never achieve consciousness.
It might even be that an unembodied AI will never achieve consciousness - that is, that direct sensory input and the ability to move around in the physical universe is critical for consciousness.
But never? No AI, no matter what approach it's built on? I'm an AI skeptic, and even I am not quite willing to go that far.
I kind of agree with rbanffy - we need a testable definition, and we don't have one. Still, we can look at AI and recognize that something is missing, even though we can't define what the "something" is.
The article never substantially mentions the subject declared in the title, and conflates it with Intelligence - whihc is something completely different.
Something, I must point out, is also lacking a testable definition.
> [AI] was conceived, if memory serves, at Carnegie Mellon University in the 1970s
I am still for Dartmouth '56. What is the author pointing to?
The author seems to be extremely abstract about the stated point, and seems to be instead interested about building up a narrative about economic control.
Until
We
Have
A
Testable
Definition
For
Consciousness
This
Discussion
Is
Meaningless
Thanks for coming to my TED talk.
I would say that LLM-based AI will never achieve consciousness.
It might even be that an unembodied AI will never achieve consciousness - that is, that direct sensory input and the ability to move around in the physical universe is critical for consciousness.
But never? No AI, no matter what approach it's built on? I'm an AI skeptic, and even I am not quite willing to go that far.
I kind of agree with rbanffy - we need a testable definition, and we don't have one. Still, we can look at AI and recognize that something is missing, even though we can't define what the "something" is.