No,Ben, you are the one wacking off an answer without any real thought.

Do you really think I was suggesting that a robot that can improvise a route through an unfamiliar configuration of boxes can automatically improvise a solution to the problem of unifying the fundamental forces in physics? That's what you're claiming I'm doing. Why? Because to quote myself:

"you're not really listening
and just automatically trying to bat away every new idea"

Your sole objective is to deny that I'm saying something new - you couldn't give a shit whether it's new or not. You have to protect your position at all costs. And that makes you a prize fool here.

If you can improvise - you can freely combine and innovate movements of different limbs. So if a robot can improvise climbing over and lifting boxes, then you can briefly show (as v. distinct from programming) it how to lift rocks and make a rock wall, and it will be able to automatically start improvising that - and you can similarly show it how to walk up mountain sides ... or climb up walls, and it can start improvising those. The power of improvisation = "take-off"

it can learn and advance by improvising just as infants do - from one physically related, "adjacent" skill and activity to another

Infants improvise everything from rolling over to sitting up, crawling, standing, taking first steps, toddling, playing with blocks - not by following precise journeys predetermined by genetic algorithms. Improvisation is the secret of their general learning powers.

But no, it does not mean they can immediately proceed from putting blocks on top of each other to solving equations in particle physics. *Eventually* they can - but it takes intermediate steps - intermediate skills to do that.

One thing we have established here - I am definitely occupying a new position that you and other AGI-ers do *not* agree with, and have not said before. And I am definitely occupying an opposite position.

You will find that improvisation is indeed the secret of general learning and skill acquisition - and the only way to journey through new fields. And every activity conducted in the real world is journeying through a new field in one form or other.






-----Original Message----- From: Ben Goertzel
Sent: Sunday, September 30, 2012 2:16 PM
To: AGI
Subject: Re: [agi] Does Siri + Watson = AGI ?

A true improvisational musical robot wouldn't just combine predefined chords
etc, it would adventurously play around with notes and create new,
surprising phrasings and chords and combinations thereof -

yeah, my  music composition program did that ;)

and might try to
invent and incorporate new instruments and sounds, just as pop music does.

but not that...

No, what I am saying is still new - and it is the secret of generality. If
you can improvise a solution to one kind of problem, you can improvise a
solution to any kind.

That's BS

A crow can improvise a solution to getting food out of a small space
-- but it can't improvise a solution to a physics equation

A person with Aspergers may be able to improvise a wonderful solution
to proving a math theorem or composing a sonata, yet may not be able
to improvise a solution to seducing the girl next door...

The Asperger's guy's neighbor may be able to improvise a strategy to
seduce nearly any woman, but when you put him in front of a musical
instrument, he may have negligible improvisational ability, just
tweaking boring pop tropes...

Your view of AGI is basically an anti-mathematician's version of
Marcus Hutter's AIXI ---- infinitely good at improvising and creating
and learning and generalizing in all domains....  That's not reality,
that's your mental wack-off fantasy, my good sir...


ben g


-------------------------------------------
AGI
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/6952829-59a2eca5
Modify Your Subscription: https://www.listbox.com/member/?&; Powered by Listbox: http://www.listbox.com


-------------------------------------------
AGI
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-c97d2393
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-2484a968
Powered by Listbox: http://www.listbox.com

Reply via email to