I gave a talk titled “The Biology of Intelligence” on Sunday to the Atheists of Florida. It went OK, I think, with some of the familiar problems of Zoom talks. I sound a bit like a Dalek at times, and early on I had a catastrophic crash — my computer went down hard and I had to restart it all — but we muddled through.
I’m not a fan of the concept of “intelligence” — it’s something we can’t define and can’t even objectively measure, yet we seem to be comfortable saying that one species is more intelligent than another. A lot of the biology research seems to be contrived towards finding morphological correlates that align with our intuition, which is just plain bad science. Maybe I just don’t trust it because nobody ever intuits that spiders are the smartest species on earth.
…nobody ever intuits that spiders are the smartest species on earth.
Nonsense, crows are smarter by far.
It is certainly true that definitions for intelligence are varied and confusing. However, what they all have in common is the ability to gather, assess and apply information to facilitate achieving one’s goals or improving one’s welfare. In this sense, an ant hill or a bee hive is intelligent, while a typical Republican voter is not.
I think the biggest problem people have is that they try to make the definition too specific, and usually too self-serving. There are many types of intelligence–even many types of human intelligence–and simple comparative descriptions of a multivariate quantities are always problematic. What interests me is the combination of intelligence types for multiple entities. Humans are great hunters. Dogs are good hunters in their own right. Combine them, and they become a great hunting team.
The same is coming to be true for machine intelligence. There are tasks a human brain does much better than a machine. There are tasks a machine can complete in an instant that would take humans a generation. The goal of AI should not be a machine that thinks like a human. Humans already think like humans. What we need are machines that think like we cannot and so extend our capabilities and build on synergy.
I suppose some day neuroscience will advance to the point that information theory can be applied to measure the relative quality and quantity of information that a brain can process over time, thus proving that Ted Cruz is stupid.
Intelligence strikes me as an attempt to push a whole range of complex phenomena into a single meaningless linear variable – usually as an exercise in gratutious one upmanship.
With cars you could have a CQI (Car Quality Index). This assesses all makes of cars on a linear scale. The Porsche 911 has a CQI of 20, a VW transporter has a CQI of 12. Therefore the Porsche is fundamentally better than the VW – unless you happen to want to move a wardrobe.
Is intelligence context-free? I.e: is a horse less intelligent than a human because they suck at math, or is a human less intelligent than a horse because they can’t look at a field of grass and see what spots are most nutritious? Is an octopus more intelligent because it does not burn fossil fuels?
I’m being facetious, but it seems to me that many of our attempts to define intelligence are very human-centric because, I guess, we’re the ones trying to measure it.
Describing an attribute in terms of its fitness (essentially evolutionary) is problematic, to me at least, since the conditions defining the purpose change. For instance, muscle strength is muscle strength regardless of whether it is useful.
Or a super computer is more powerful than a calculator, even though a calculator may be perfectly necessary and sufficient for it’s purpose.
OTOH, there’s the SF trope wondering if high intelligence leads to self destruction, so at some point does intelligence = stupidity?
(I haven’t had a chance to watch the video, so apologies if any of this is already answered.)
Stroppy @ 6
Since intelligence permits the spread of cultural traits that are cumulatively harmful (bad memes, like the inherent superiority of the master race, or war as
a cleansing process that ennobles the people ) it has extra failure modes.
But short term, intelligence is definitely useful once a rich material culture has developed. The Inuits do not freeze to death during winter.
birgerjohansson@#7:
The Inuits do not freeze to death during winter.
That is not true. There were a few, but they were republicans and insisted in walking about in shorts and flip-flops because it expressed their freedoms.
Spiders are obviously more intelligent because they live in houses.
I suppose when we say intelligence, we are only considering the intelligence that exists in our conscious minds without including the intelligence used to do things like throw a dart accurately or catch a ball. In my case though, my best ideas / insights seem to just pop into my head. I can write a program, but the kernel of the algorithm often seems to come from nowhere, and I just use my conscious mind to write the program neatly and clearly.
Since intelligence permits the spread of cultural traits that are cumulatively harmful …
What if intelligence is a cultural trait? I mean, we accept (except for a few scientific racists) that intelligence is largely culturally determined, right? What if there are a variety of strategies for cognition and most of us learn the default ones and the people we consider highly intelligent are using modes that are better for some contexts like test-taking? For example, those of you who have studied how to crush an SAT know there are specific strategies for multiple choice tests. It’s not that the subject is more intelligent (whatever that is) but simply that they learned a better way. Someone who uses a backhoe to dig a ditch isn’t stronger than a guy who uses a shovel, either.
I know it’s popular to adopt the position (as I do) that “intelligence” is partly learned and partly innate but it’s very hard to figure out the proportion. What of it’s all learned? Humans are born cognitively helpless and studying what infants have “built in”, as Piaget did, shows that pretty much everything, including tracking moving objects with our eyes and object persistence are learned. I think it could be hard to demonstrate anything about intelligence that is innate given the dominance of learned strategies.
[I am not arguing with you, just trampolining off your comment]
What the machines need are people that think like they cannot and so extend their capabilities and build on synergy.
Inuit is plural. Inuk is the singular.
Inuit. My bad.
.
Intelligence: we are ‘programmed’ to learn, and learn quickly , but what we learn is cultural.
Being a mesolithic hunter-gatherer would not have been simple, a big brain is ‘expensive’ so we have the smallest brains we can get away with whlle existing (and competing with other humans).
And then culture started stacking ideas and innovations on top of each other. We got antibiotics, nukes, fascism, humanitarianism, scientology, stamp collecting and -maybe, someday- a cure for cancer.
Why not say instead that intelligence is learning? Not a thing to be learned — because there isn’t anything in the world corresponding to something like that which we’d identify as “intelligence” — but an ability to go through a process of learning.
I mean, it’s not about a set of facts dumped into your brain which is able to store them. That is a different sort of concept, which is closer to how we think of memory or knowledge. (And just leave aside how crude that brief description is — as inaccurate as it may be, that is roughly the phenomenology that people have when they think of them in those terms.)
When you’re acting intelligently, you’re actively adjusting to new information/stimuli, making inferences about these things, and connecting them together in a way that motivates a response that will be appropriate/useful (given the situation you happen to be in). You could dress it up in various ways to be a little more sophisticated, but this is more or less the basic idea.
Anyway, some species do have more of a capacity to do that sort of thing than others. I think that’s fairly obvious and uncontroversial. I don’t think we have to reject something like that, on the grounds that it’s a human-defined concept which is driven by the kinds of things humans tend to value. I mean, so what if it is? It’s okay that organisms in a different species could have different ways of evaluating and categorizing things based on their own perspectives and abilities and whatnot; that doesn’t actually invalidate ours. They’re just different, that’s all.
@5 Marcus,
I agree. I recall the Voyager probe with the “disc” containing information that, supposedly, an intelligent species would be able to decipher. One of the images is an arrow that points to the third planet. Obvious to us. But, what about a species that never had arrows? That’s a concept unique to our development. I think one could argue that all of the symbols are unique to us and, therefore, bewildering to another species, if, in fact, another species has bewilderment.
Doc Bill,
It’s inferable — definitely so if the possibility space can be exhaustively explored and the best consilience found.
Marcus Ranum @ 8
Convergent evolution: Republicans have evolved several times, but for some reason they keep going extinct.
.
“Spiders are the smartest ”
(cough) Portia?
Reasoning, problem solving; making tools is kind of meta, some animals do it, but as far as I know only humans make tools in order to make other tools, and then can use them to answer questions beyond just survival; telescopes for instance–and computers, which are more than just extensions of physical powers, but extensions of conceptual powers. Meta-meta.
It’s almost as though we can consciously rebuild our own brains… So would genetically engineered, cybernetically enhanced additions to a human count as part of innate human intelligence or something else?
I don’t know about dolphins though, if they had hands who knows what? And gorillas, probably a lot closer to us than we think but for lack of dexterity.
And then there’s mice, and perhaps spiders, which are pandimensional, hyper-intelligent, three dimensional projections of beings whose actual forms and understanding we can’t possibly comprehend.