On 2024-10-03, faeychild <
faeychild@nomail.afraid.org> wrote:
On 3/10/24 20:11, Daniel70 wrote:
... and the problem with A.I., as I see it, would seem to be that it may
be 'A.' (Artificial) but I have great doubts about the 'I.'
(Intelligence)!!
Maybe, one day in the future, 'A.I.' WILL be 'I'!!
Agree! I haven't bothered much with A.I.
How exactly does it work?
It can't be just a rote process. A huge list of possibilities with an
It is a "rote process" but also probablistic. It goes through the web,
and finds correlation between the sysmbols (letters, phonemes, sysmbols, words....-- apparently Google now uses more than million of them)
including how far apart they are the texts it uses for training (so that
gives a 10^12 times distance matrix of occurances, then runs it though a transfomation which reduces that to a manageable number then runs your
question throught that and makes up an answer by probabilistically
choosing sybmbols
Remember that training includes all of Trumps speaches which are on the
web.
It is a sociopath. It does not care about truth, or any other moral
quality you might happen to think about. (The people working in the area
call those sociopathic answers "halucinations" as if the program had a
concept of reality.
"If Then Do This But" script.
A dog is far more intelligent. Brilliant comparatively.
Depends on in what aspect.
--- MBSE BBS v1.1.0 (Linux-x86_64)
* Origin: A noiseless patient Spider (2:250/1@fidonet)