[LINK] Machine Learning Was: Re: Robot cars and the fear gap

[David Lochrin]

On 2016-07-28 11:07 Karl Auer wrote:

> On Thu, 2016-07-28 at 10:25 +1000, David Lochrin wrote:
>> philosopher John Searle developed an argument against Strong AI known as the "Chinese Room" thought experiment to which I referred earlier, [...]  However it has stood up well against attempts to knock it off.
> 
> Douglas Hofstadter comprehensively demolished it, and I think Daniel Dennett.
> 
> It's much more a rhetorical treatise than a philosophical one. Lots of word games.

I see Hofstadter discusses it in Metamagical Themas (Ch. 26, p.631 in my copy) and I'll read it later for interest.


>> There's no colour in physics, only EM waves of certain wavelengths or photons of certain energies, so where would it come from?  If you can answer that you'll be famous.
> 
> Define "perceive" :-) Then prove - or even demonstrate - that you (or anyone) is doing it, and how you (or anyone) doing it is any different from a machine that does it. I'll wait over here.

The main point I'm making is that nobody understands the nature of consciousness, even Douglas Hofstatder.  The enthusiasts who have such touching faith in "strong AI" are relying on just that, faith, and they're now the second or third generation to do so.

Part of the problem may be that we're using our brains to investigate our brains.  Everyone knows "perception" in an internalised way but it's not something which can be objectively observed.  And that was the point of my earlier reply to Jan about the electronic analogue of a brain.  On one hand we'd expect it to behave the same as a biological brain, why wouldn't it?  But on the other, physics has no scope for perception to arise.

Hofstatder states "To me, the major question in AI is this: "What in the world is going on to enable you to convert from 100,000,000 retinal dots into one single word 'mother' in one tenth of a second?"  Perception is where it's at!"  I agree.


> Your statement that "the output of a fibre can be considered a symbol" might need just a leeetle more scaffolding, too. Even if "in the sense of information theory."

Yes, it wasn't intended to be deconstructed too much!  But I'm afraid I don't have time to go there just at the moment.

Regards,
David L.