Re: [Sheflug] Royal Institution Christmas Lecture

["Alex Hudson" <nospam [at] alexhudson.com>]

> > Go on then, show me a definition of machine learning that specifies what
> > implementations you can and can't use?
>
> Warwick's words:
> "These robots learn from each other"
> Reality:
> They don't. They can't. How can they? They can't even communicate with
> each other, they're based on a simple GA.

Whether or not you're disputing that they learn wasn't my point; you said
Warwick said they learnt, but they were made using implementation X. That
they are based on a certain implementation doesn't prohibit their learning.

> He's a liar, and IMO very dangerous to AI and cybernetics research.

You just disagree over what learning is. People also disagree over
intelligence, hence arguments over even the Turing test. You say, "They
don't learn". Warwick obviously believes they do. Learning doesn't require
communication, in fact there is very little learning does require. If you
and Warwick agreed on what 'learning' meant (and I predict that you probably
don't ;) then the basis for you saying they don't learn would be obvious and
testable; saying they don't learn because of reason A, and because they use
implementation X is an invalid argument.

BTW, check out ntk.net - quoteth I:

    "One of the ideas mooted for the webcast of this year's Royal
Institution Christmas Lectures involved affixing a wireless "KevinKam" to
the side of the man-machine symbiont's head."

Read the whole article / paragraph. C'est tres amusant, c'est need to know!

Cheers,

Alex.

---------------------------------------------------------------------
Sheffield Linux User's Group - http://www.sheflug.co.uk
To unsubscribe from this list send mail to
- <sheflug-request [at] vuw.ac.nz> - with the word 
 "unsubscribe" in the body of the message. 

  GNU the choice of a complete generation.