Moravec's paradox

#1

Post by subnaught » Sat, Feb 06, 2016, 6:21 PM UTC

Re: Moravec's paradox

#2

Post by danrisso3 » Sat, Feb 06, 2016, 8:13 PM UTC

People who think that high-level reasoning is easy compared to sensorimotor skills have successfully implemented neither and lack both.

-- Benjamin Franklin

Neural networks can be used to balance robots. They can't be used to generate Agda proofs of non-trivial theorems, or C programs that do something useful, etc.

Re: Moravec's paradox

#3

Post by adaml_623 » Sat, Feb 06, 2016, 8:59 PM UTC

danrisso3 wrote: post #2

People who think that high-level reasoning is easy compared to sensorimotor skills have successfully implemented neither and lack both. -- Benjamin Franklin Neural networks can be used to balance robots. They can't be used to generate Agda proofs of non-trivial theorems, or C programs that do something useful, etc.

Thank you I was trying to remember who said that quote ;-) ...

But seriously I'm not sure that just because checkers and various logical puzzles lend themselves to simple algorithms equates to "high-level reasoning requires very little computation"

I mean unless I've managed to miss all the AI bots sitting n the web asking me to talk. I think this might be a paradox that only existed until people thought more carefully about high level reasoning.

Re: Moravec's paradox

#4

Post by mcphage » Sun, Feb 07, 2016, 4:05 AM UTC

danrisso3 wrote: post #2

People who think that high-level reasoning is easy compared to sensorimotor skills have successfully implemented neither and lack both. -- Benjamin Franklin Neural networks can be used to balance robots. They can't be used to generate Agda proofs of non-trivial theorems, or C programs that do something useful, etc.

> balance robots

Making a robot that doesn't fall down is a really low standard.

Re: Moravec's paradox

#5

Post by phamilton » Sun, Feb 07, 2016, 6:00 AM UTC

This is a decent motivation for "manual automation". Figuring out what is a face in a picture is hard for a computer but easy for a person. Comparing that face with thousands of existing faces to eliminate 99% that aren't anywhere close to a match is easier for a computer than it is a person. Deciding if it is a true match is easier for a person than a computer.

There are a lot of situations that would benefit from such a workflow.