A.I. Developer Challenges Pro-Human Bias

destinyland writes "After 13 years, the creator of the Noble Ape cognitive simulation says he's learned two things about artificial intelligence. 'Survival is a far better metric of intelligence than replicating human intelligence,' and "There are a number of examples of vastly more intelligent systems (in terms of survival) than human intelligence." Both Apple and Intel have used his simulation as a processor metric, but now Tom Barbalet argues its insights could be broadly applied to real life. His examples of durable non-human systems? The legal system, the health care system, and even the internet, where individual humans are simply the 'passive maintaining agents,' and the systems can't be conquered without a human onslaught that's several magnitudes larger."

Comments

Bad metric

[dgatwood]

Survival is a terrible metric of intelligence. By that standard, lions and tigers and bears are the most intelligent species on the planet.

Re:Bad metric

[MrMista_B]

You mean stupid. Most lions and tigers are endangered, if not close to extinction, and bears aren't too well off either.

A better example would be insects, like mosquitoes.

Re:Bad metric

[johnsonav]

Well, if your aim is to develop artificial intelligence, intelligence is probably a pretty good metric to determine how well you've performed the task you set out.

Well, that seems a little too easy. Now all we need is a definition of "intelligence" we can all agree on...

Re:Bad metric

[Hurricane78]

Depends on what you define as "intelligent".

Survival is the metric for success. And if you are the one surviving, you define what "intelligent" means.

Try doubting it from your grave. ^^

And (our) insect( overlord)s by far rule this world. Their only problem: They don't know what "define" means. ;)

Re:Bad metric

[johnsonav]

Bingo. Intelligence is one of those corporate feelgood words, like state-of-the-art, or user-friendly. They are completely impossible to quantify.

Exactly. That's how we ended up with things like the Turing Test. I can't define intelligence, but I know it when I see it.

But, that leads to the problem of a human-centric view of intelligence. We have such a hard time defining human intelligence, defining non-human intelligence will be almost impossible.

Re:Bad metric

[Kjella]

I think we've defined intelligence almost like the halting problem, it's everything can't be solved by an algorithm and every time we find something solved by an algorithm we exclude it. Every time computers and robots do something we reduce it to mere execution of an algorithm, even when the algorithm wasn't defined by a human like in neural nets. As long as it stays within the problem domain we'll never consider it intelligent, intelligence is creativity and thinking outside the box. The best sign of intelligence in a Chess program would be "Want to play a game of Go instead?"

He's too close.

[Toonol]

By redefining intelligence to have nothing to do with what anybody means by intelligence, he can then claim that other systems exhibit more intelligence. Like a rock, presumably, since it survives far better than humans. I think this may be an example of somebody getting too interesting in specifics of tree-bark, and forgetting about the forest.

Re:He's too close.

[Trepidity]

This seems to be a common mode of argument for people who for some reason don't like what people commonly mean by "intelligence", which is something closer to "critical thinking skills combined with ability to acquire, retain, and use information", but nonetheless like the aura of the term. There's been a decades-long wave of politically correct attempts to broaden intelligence to include other things, like "emotional intelligence", which might indeed be important, useful, and worthy of study, but aren't really what the word "intelligence" means, so should probably get new names instead of being shoehorned in there. Now we've got survivability, which is indeed an interesting trait of an organism, but is not in itself actually what anyone calls intelligence (though being more intelligent might help with survivability, at least in some contexts).

It's a perfectly valid argument to say: look, I don't think intelligence is the most interesting property to study; here's this other property, which might overlap somewhat, but I argue is more interesting. But pretending that your new property is really intelligence is a weird sort of linguistic move, because your property is not what people use that word to mean.

Re:He's too close.

[rlseaman]

There's been a decades-long wave of politically correct attempts to broaden intelligence to include other things, like "emotional intelligence", which might indeed be important, useful, and worthy of study, but aren't really what the word "intelligence" means

Point taken, but you are confounding two separate issues yourself. The notion of Howard Gardner's so-called "multiple intelligences" is well presented in Stephen Jay Gould's book, "Mismeasure of Man". Gould's thesis is that IQ is a meaningless measure, and that intelligence is a meaningless notion that doesn't correspond to a single measurable entity in the first place.

You suggest a definition "critical thinking skills combined with ability to acquire, retain, and use information", but this begs the question by assuming its own premises. In the first place, you describe a composite entity comprising multiple skills (there's Gardner's multiple intelligences) as well as something ("ability to acquire, retain, and use information") that seems itself like a circular definition.

So yes, there is a bit of academic slight of hand in reusing the word "intelligence" to represent something other than "what people commonly mean", but the fundamental point is that what people commonly are trying to express is a bunch of hooey.

That said, this statement from the referenced article: "survival is a far better metric of intelligence than replicating human intelligence" seems evolutionarily extremely suspect. Survival is the dependent variable in Natural Selection. Phenotypical traits like intelligence, whether multiple or singular, are the independent variables driving evolution.

Re:He's too close.

[Trepidity]

I don't see it as some sort of prerequisite for a word that describes humans to describe a single entity that's empirically testable. People use phrases like "kind" and "loving" and "artistic" and "creative" to describe humans, even though there is probably no solid definition that's empirically testable. I'd still resist some scientist trying to take one of those terms and apply it to their own pet concept that happens to be empirically testable but isn't what the word actually means. Inventing new jargon, while less sexy, would be less confusing.

I prefer a more human metric.

[FlyingBishop]

He essentially seems to be arguing that grey goo is the pinnacle of AI.

I much prefer the existing literature requiring that intelligence be an intelligence we can relate to as humans. Survivability is an interesting metric for creating more self-sustaining systems, but the goal of robotics should be fostering better knowledge and understanding of the universe. Searching for blind replication at the best rate possible just feels empty.

Re:Banks

[johnsonav]

The banking system is another example of a system much better than human intelligence for survival and resilience. Oh wait...

It persuaded us to save its "life", didn't it?

Relativism strikes again...

[Twinbee]

Uh oh, it's one of those hard-line relativist type rants again.

One choice quote from the article:

The same reason you get the opinion "The primacy of human intelligence is one of the last and greatest myths of the anthropomorphic divide

Okay, human intelligence may be fuzzy and difficult to objectively measure. But that applies to many things such as CPU speed, Kolmogorov complexity, how complicated a shape is, or how much heat/sound insulation a particular material provides. Even how good a piece of music/art is.

They're tricky, but there's no doubt that exponentially low and high numbers can be given to each of those attributes.

Re:Relativism strikes again...

[Twinbee]

And now listen to some of the latest Japanese pop - don't you find it just a tiny bit odd how much like Western pop it is? (I enjoy all types of music, and think most (but not all) modern pop is crap for the record).

Look I'm not saying I have the best taste in music in the world, and I've known people who at least partially subscribe to relativism and have *decent* taste in music. I've known the reverse too (people like myself, except with probably bad taste).

But let's not start putting all music (or even a culture's music) on an equal footing shall we. Every time you hear a piece where you can't possibly see what's good in it, start to think that it may not be the music intrinsically that people are enjoying, but rather the indirect feelings and associations they are getting, as a BYPRODUCT of that music.

For that reason and its historical role as the refutation / logical conclusion to serialism, yes it is a masterpiece of composition.

So it's great for all the wrong reasons. Intrinsically (which is what should count), the music is poor, because it would never stand the test of time. People in 500 years time won't start saying, how 'wonderful' it is. Because they can't (it's garbage). The most they can say is how much it affected society. That may be notable in its own right, but *please* don't go as far as to say the music is intrinsically of worth.

Oh and by the way, complexity may not be the ultimate goal, because many great pieces are simple, but I would say it has a higher potential of being good. The more complex a piece of music, the more difficult it is to make it good too, but the rewards are better.

I'm in the UK by the way.

Re:A bit of a Summary

[Bongo]

So here's what he says: A system, such as the health care or legal system, will not be shutdown by one person. In fact, it probably won't even be shutdown by 10 people, maybe 100. And hence, the system is vastly more intelligent than a human, intrinsically since we worked in numbers to evolve this system.

In philosophy (couple of books) there is a discussion about how various fields confuse individuals and systems. Like, Nature is a huge complex system, and man wouldn't survive without Nature, therefore Nature and the ecosystem are more important than Man. Therefore man is just another species, and man must learn his place and minimize his impact. Well, there is some truth to that, but the underlying confusion is that they're comparing an individual organism with a massive complex system, under the guise that the organism is just another complex system anyway. Similar confusions come up when people talk about whether society or the individual is more important. I had one Marxist tell me that I am "nothing" without society. Well, again there is some truth to that, but it is only partially true.

It is not just that we don't like comparing ourselves to other more complex things, and feel uncomfortable about it. It is that these different things have some very different properties. An individual organism like a person has sentience and self-directed intentionality. Society doesn't have sentience (at most it exhibits "flocking" type behaviors) and an ecosystem doesn't have sentience (despite what some new agers claim about the planet being "conscious").

And meanwhile, society has properties that can't be reduced to individual consciousness. We have the English Language, and you have to be born into or join a society of English speakers in order to learn it. We have ethical codes, which again are about social interactions. If I was the only person on the planet, the only being, there would be no need for ethics. They wouldn't exist without some sort of collective to bounce good and bad off of. And these social structures do indeed "last longer" than individuals, and can't be torn down by individuals, not because they are more intelligent, but simply because they exist in a different domain to the individual. They are a different side of the coin. They are distinct but related to the individual.

But also notice, that without individual minds interacting with each other, there would be no social system, no legal frameworks, no ethical codes. Just like you can't have an ecosystem without organisms interacting. And as everyone here is saying, if you start to mis-assign a quality that belongs to one domain (sentience, intentionality, intelligence) to a different domain (ecosystems, legal systems) you end up in weird and wrong places (but its research so who knows what might come of it).

But it does end up looking like, because modeling human intelligence is so hard, we'll just change fields and start modeling systems instead, and you know, maybe we'll get somewhere with that, and nobody will notice we just changed our research area.