On 2023-01-17 4:52 p.m., Chuck Guzis via cctalk wrote:
On 1/17/23 15:34, Bill Gunshannon via cctalk wrote:
On 1/17/23 12:10, Chris via cctalk wrote:
It seems at this point that AI can only look shit up.
Which is all Watson could do when it "beat" humans at Jeopardy.
I don't doubt that eventually these things will learn how to reason
to whatever degree.
Science still has no real idea how we think. Or how we learn. Or much
else about the actual functioning of the brain. Until they do how can
anyone think they can teach a computer, that real is as dumb as a rock
My point in selecting the IBM 7080 autocoder was just this. There are
almost no 7080 autocoder programs published on the web, but there are
reference manuals (bitsavers). Is an AI smart enough to go the human
route and *read the manual* and draw conclusions from it sufficient to
write a 7080 autocoder program?
If not, the AI is just a fancy regurgitation machine.
--Chuck
And now we know why they all go mad.
I still think HAL 9000 was the way to go,emulate a analog system, learn
things slowly,grow, and use 'brain' like sub systems.
How many people in AI,read
The Computer and the Brain from 1958?
https://archive.org/details/computerbrain0000vonn
My 2 cents.
Ben.
