Re: Intelligent Evolution

"J.A. Legris" <jalegris@xxxxxxxxxxxx> wrote:
On Aug 10, 5:30 am, c...@xxxxxxxx (Curt Welch) wrote:

The thing you need to explain in greater detail, which is common among
crackpots, seers, and other loons is how your claims might be be
falsified. By your criteria, all states that persist, including those
that persist in not existing, do so by virtue of feedback from the
environment (i.e. reinforcement) that contributes to the persistence
of those states. To test such a claim, we would have interrupt all
feedback from the environment, which is not going to happen in this
universe. Of course, this provides a neat explanation for the Big
Bang, which is that if you have nothing, including no reinforcers to
enforce that nothingness, then you must have everything. Boom!

Well, what I'm talking about here, looking at evolution as a process of
reinforcement, is not something that needs to be falsified. If you simply
look at it right, you see the parallels I'm talking about. There's nothing
here to prove. It's like saying: "if you look at a cylinder from the side,
you see a square". If I pointed that out to you, would you then turn to me
and say, how can you prove what you say is true? If you can't falsify what
you suggest, then it's just the rambling of a crackpot?

The parallels between reinforcement learning and evolution are there for
anyone to see who wants to open their eyes and look at it. It's something
anyone can verify for themselves just by looking at it and thinking about
it.

If you don't understand what the parallels are, or can't see them for
yourself, I'll be happy to talk about them in more detail so you can see
them, which is what I was offering to do for John. However, he seems not
to be interested in that. Such high level abstractions don't seem to be of
much use or interest to him.

If by "your claims" you were making reference to my never ending insistence
that human behavior is simply the behavior of a large general purpose
reinforcement learning machine, then there's really nothing I can prove by
falsification. At least I can't think of any way to do that. The idea is
so simple, that it can't be falsified. In other words, I simply can't
prove it correct that way. All I can do, is continue to look for a working
algorithm (one that I could explain to John and have him code it), which
produces the life-like behavior I expect these algorithms to produce.
Then, with your own eyes, you can look at a robot programmed with such an
algorithm, and see for yourself what I've been talking about. Until I can
produce that working demonstration, I've got nothing I can prove. All I
can do, is point to all the clues that are consistent with this idea to
make more people give the idea some serious considerations on their own.

Working hardware which produces behavior that the majority of people can
agree is more life-like than anything done by AI researches in the past, is
the only real proof I know of.

--
Curt Welch http://CurtWelch.Com/
curt@xxxxxxxx http://NewsReader.Com/
.

Relevant Pages

  • Re: Intelligent Evolution
    ... among crackpots, seers, and other loons is how your claims might be ... reinforcement, is not something that needs to be falsified. ... If you can't falsify what you suggest, ... look at a robot programmed with such an algorithm, ...
    (comp.ai.philosophy)
  • Re: Intelligent Evolution
    ... crackpots, seers, and other loons is how your claims might be be ... reinforcement, is not something that needs to be falsified. ... If you can't falsify what ... algorithm, and see for yourself what I've been talking about. ...
    (comp.ai.philosophy)
  • Is evolution an intelligent process?
    ... Subject: Intelligent Evolution ... evolution as a process of reinforcement, ... can't falsify what you suggest, ... reinforcement learning machine, then there's really ...
    (comp.ai.philosophy)