Re: Emotion Machine Critics
- Marvin:However, the idea of adding "reasons" for these goes much further,
because this makes it possible to think about making higher-level,
more powerful models, which then can make new kinds of prediction
that are not (simply) based on experience. Your Dennett quotation
does not go on to discuss more reflective systems that can think and
reason about its models. It looks like you (smartly) added that!I take your point - I think. And Dennett does indeed as I said have a higher "Gregorian" level of intelligent evolution. But what this exchange usefully highlights is the enormous ambiguity - and many different levels - of the idea of an agent "reasoning" about its mistakes, and altering its models.There can be no doubt, for example, that crows, say, and beavers along with many other animals reason consciously and nonverbally about their mistakes and alter their strategies - the one developing relatively complex hooks, the other developing new ways of shoring up dams - through patient trial and error. Come to think about it, I'd be surprised, given Darwin's description of their complex burrow-filling & building activities, if worms don't reason and correct themselves consciously.And infants must reason preverbally about their mistakes.And then there are levels upon levels of adolescent-to-adult reasoning and correction of models. For example, only the intellectual, I guess, will be able to reason with confidence about the various paradigms and schools of thought of different sciences and disciplines. That requires a very high level of abstraction and generalisation.Your model of intelligence is non-evolutionary and non-developmental - and my point is: that just won't do anymore. And it won't work for AGI - it's like trying to build a tower (Dennett talks of an evolutionary tower of "generate-and-test") - starting in mid-air!
- --- In firstname.lastname@example.org, "anssihyytiainen" <anssih@...>
> --- In email@example.com, "feedbackdroids"
> <feedbackdroids@> wrote:
> > So, parrots, dolphins, and sea lions aren't good enough. Would
> > REALLY want a talking cat? It would do nothing but whine. Feedme,
> > feed me, etc, echoing ....cats.
> Heh, so I would imagine :) There's definitely a market for whining
> Anyway, this reminds me of this old video:
> The first two are doing fairly good job at parroting, but the last
> seem to be just making weird noises of their own...cats
> What would be interesting is to find out what sorts of sentences
> (and dogs) could learn to form. I would expect the sophisticationof
> the language is directly proportional to the intelligence level ofthe
> cat, and there are definitely differences between individual cats.I
> would imagine that some careful breeding should also do wonders. Iuniquely
> guess we are so used to think of our own language abilities as
> human, that no one has seriously thought of breeding talkinganimals :)
> (Or has there actually been attempts like this?)Despite your arguments above, I suspect no one on their right mind,
knowing how cats presently love to whine, would choose that
particular animal to try and evolve into true talking.
[by way of example, last night as I was walking home, a tabby jumped
out of the bushes in front of me on the sidewalk, and
said "Raaaaaaaa!". The usual word for "pet me, I need some