To follow this blog by email, give your address here...

Sunday, April 26, 2009

Why AGI Researchers Should Care about Cetaceans

This is just a brief follow-up to my last post, and a prelude to the one that will follow, which is already brewing in my brain....

In case some of y'all are wondering why I ... whose main intellectual obsession is the creation of AGI systems with general intelligence at the human level and beyond ... have suddenly started ranting about cetacean intelligence, I suppose I should be more explicit about my research-related motivation for digging into the topic

Of course there's a personal motivation -- I love nonhuman animals ... at the moment as well as some humans I share my house with a parrot, 2 dogs and 5 bunnies; and there have been friends of a lot of other species in my life at various times.... In fact the parrot named Abaca might be classified as my best friend over the last few months ;-) ....

I've encountered wild dolphins up close in the shallows of the Indian Ocean when I lived in Western Australia 13-15 years ago, and was certainly struck by the experience, as brief and superficial as it was. I definitely wanted more (and would have sought out more if I hadn't left Western Oz to move to New York and start an AI company)

But what I want to focus on here is my intellectual motivation for, as an AGI researcher, finding cetacean intelligence important.

How often do I hear, among AGI researchers, words to the effect of "Of course we need to model our AGI systems on the human brain and mind, since after all it's the only example we have of a highly generally intelligent system."

I tend to resist this line of thinking ... I think we understand enough about the general mathematics and computer science of cognition that we can understand general intelligence in a manner going beyond the human-specific. My own AI work is an amalgam of aspects directly inspired by human intelligence, and aspects inspired by a broader understanding of intelligence.

But still, there is some point to the common observation that we only know one example of a highly generally intelligent system: the human brain.

But is it actually true?

What if the subset of cetacean intelligence researchers who believe cetaceans have general intelligence comparable to, or greater than, human intelligence ... are actually correct? (Which is my suspicion.)

Then in fact there's another example available -- and we're just not taking the trouble to study it as carefully and thoroughly as we should.

In my immediately previous blog post I gave some links into the cetacean intelligence literature, and some speculations as to what I think the broad nature of cetacean intelligence might possibly be.

In my next blog post I'll discuss some cutting-edge approaches that we might take over the next couple decades to more thoroughly understand cetacean intelligence.

I'm not suggesting that resources be taken from AGI and redirected to cetacean cognitive science: I think that both areas are distressingly underfunded.

In the case that we create AGI programs with superhuman general intelligence before we understand cetacean minds, I think we might still have something to learn from the minds of dolphins. Because cetacean minds may possess a quite different form of intelligence than either us or our AGI creations. And it's hard to tell what may be learned by studying some advanced, fundamentally different incarnation of intelligence.

There could possibly be interesting implications for the study of AGI ethics here, for instance. Cetacea are certainly not optimally ethical creatures ... they're capable of violence just like most other mammals ... but based on what we can understand today, it seems their social organization may have fewer egregious ethical issues than ours. As one example, they seem to have achieved a large-scale, global social organization without warfare. (Evidence of the global nature of their social organization is tentative, but provided by observations such as the way repeated phrases in whale "song" tend to arise in one part of the globe, then spread through whales in all the world's oceans, then die out after a time, replaced by others.)

I'm certainly not suggesting that study of cetacean society will magically provide the answer to the AGI ethics problem, or the problem of generally understanding general intelligence, etc. However, I think it would be very interesting to understand how a fundamentally different sort of general intelligence works, and how it has approached the society/ethics problem, as an additional body of evidence to utilize as we shape the minds of the future.

3 comments:

Odin Khriswill said...

Even if the most intelligent of cetaceans turn out to be only half that of humans, I can still see the merit in this. In fact, knowing more about what abilities they possess and we possess along with our biological differences could provide great leads on the nature of human mental abilities. Or maybe something more unexpected.

Ben Collins said...

Ben, I think your highly intelligent mind should be concerned primarily about the Pandora's Box of technology, not the abstract possibility of other intelligences in the ocean or elsewhere.

It is an admirable endeavor to ponder and research animal minds, but in my opinion the development of technology is a fundamental component of "general intelligence" that is lacking in dolphins or any other species.

I think the idea that somehow dolphins are smarter than us is false, even if we ultimately kill ourselves with technology and dolphins survive. Eventually another species will discover technology (possibly them) and face the same kind of intelligence complexity that comes with of evolutionary and acceleration of technology.

My dog has feelings and ideas and so forth, but he's dumber than me. Dolphins are too, at least on one critical level. I realize that defining intelligence is a challenge (I read, but did not have the intelligence to fully grasp your recent draft paper on the subject), but you can't move forward with your own research that could result in the birth of a new being far more intelligent than humans or dolphins while ignoring technology and its relationship with intelligence.

We discovered technology and dolphins didn't. Maybe they are lucky, but they aren't smarter.

Ben Goertzel said...

Ben Collins: the question of "how smart are cetaceans" is much less interesting than the question of "how are they smart" ...

And of course, looking at cetaceans is not critical to my approach to AGI, which is not founded on trying to closely imitate ANY biological intelligences.

I just find it interesting to investigate general intelligence architectures besides the one we usually think about (humans).

Of course tool-building and technology are super-important to general intelligence ... but there may possibly be other aspects that are also very important to general intelligence, in which cetacea exceed us ...

It seems unlikely to me OTOH that dogs or parrots exceed humans significantly in aspects that are extremely useful for general intelligence. So I don't find them as interesting to investigate as cetaceans ... though I admit I am endlessly fascinated by the adaptable and intelligent responses of the parrot who sits behind me and keeps me company as I work.... (Her vocal responses to music, for example, are really quite subtle at times, though this sort of thing is hard to quantify...)