Standard objection: "but it has no incentive/ability to route around the hard-coded utility function" Response: if it's constrained in its thinking in a way we're not, it can hardly be called superintelligent.
-
-
Neural nets will never produce consciousness. Which means it almost certainly will never be smarter than a person.
-
pretty daring to assert the level of understanding of consciousness necessary to assume the question that it's produced by anything, much less what it can and can't be produced by
- 1 more reply
New conversation -
-
-
Relevant: are human-readable "representations" of what neural nets are doing (encoding/decoding) necessary for neural nets to do what they do? https://arxiv.org/pdf/1904.10396.pdf …
-
It's necessary that human-reading representations exist, even if we don't happen to know what they are. This is a function of Turing completeness. There's nothing a neural net can do that a human couldn't also do.
- 1 more reply
New conversation -
-
-
The successor technology probably does depend on that, and after it's on its own, I'd bet.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.