Artificial or Natural Intelligences

I wrote most of this in a comment previously, but it is important enough to put it up here. The core subject here is whether or not we can hope for human intelligence and consciousness to evolve as breathtakingly rapidly as would be necessary to keep up with the accelerating technical progress we're making, so that we'll make it in time to not destroy ourselves and the planet, or whether we have to accept that a benevolent Artificial Intelligence has to be constructed that will keep us from doing so. The latter is a very serious endeavor being pursued by some very smart people who believe it is our only hope.

First of all, it is quite clear that the human mind that most of us are sporting at this point is woefully inadequate when it comes to logical thinking. If it is up to us to figure out rationally to do the right thing, it isn't looking very good.

Michael Anissimov provided an excellent list of some of human logical fallacies that have been studied academically. A lot of relevant material can be found by looking up some of the following terms in Google: availability bias, conjunction fallacy, Wason selection task, support theory, representativeness heuristic, misperception of random sequences, expert judgement and uncertainty. Essentially they constitute a number of different ways that humans habitually let themselves be influenced by incidental irrelevant factors into making illogical conclusions. Our answers depend greatly on how the questions are framed, and are likely to contain traces of whatever random garbage we happen to have been fed just before. You know many of the more simple examples of that from simple party games, like: "Say 'top' ten times fast!!"......... "Now, quick, what do you do at a red light?" Most people will say "Stop". That's an easy one, but there are many more complex and insidious fallacies in logic, which quite easily can be domonstrated.

The human mind is having a major problem when it aspires to being logical and rational, or even sane, for that matter. The way even the most coherent of us are making decisions is, in the bigger picture of things, horribly sloppy. And it isn't even just that it is sloppy, but the faulty thinking is hardwired into the way that our minds work in the first place. And it is kind of scary to think that even our best scientists and world leaders are subject to the exact same fallacies, and there's no guarantee that they've learned to transcend most of them. A "consciousness of abstraction" a la General Semantics might include awareness of the faults into our thinking process, in order to consistently arrive at more sane results, but it still isn't quite enough.

Michael called the human mind "a pinhead-sized box in a galaxy-sized space of minds-in-general". And, yeah, compared with how smart one possibly could be, we aren't doing all that well.

But after saying that, let me emphasize that what I have confidence in, when it comes to human evolution, is not particularly an individual human ability to think oneself logically into acting in the sane and rational and responsible way that the control of our new technological capabillities would require. Rather, I postulate that there's something systemic going on, of a higher order, which possesses a much more coherent intelligence, of a different nature than that of the individual human.

I postulate that Nature effectively has evolved the equivalent of an AI way past a singularity point. Or maybe rather that it has had it all along. For that matter, I would tend to believe that life can't really happen without the presence of an intelligence that guides it.

I don't particularly mean that in the sense of some anthropomorphic God that sits and hands out arbitrary decisions. I don't believe in such an entity. I'd rather see it as an intelligent system. Not exactly an "overarching benevolent architect" although that doesn't sound as bad to me as it does to some. But I see it more as a built-in intelligence than as some external architect. Although there's always the question of how it came about in the first place, even if it is now self-sufficient.

Just like the ant doesn't exist without the anthill, I suspect that any individual life-form wouldn't really be alive unless it were the smaller unit of something a good deal smarter than itself. The Natural Intelligence that ensures the coherence and evolution of that species. Or an eco-system. Or something else I don't understand.

We humans have a tendency to suffer from hubris and think that one of us is somehow the very spearhead of evolution, the smartest thing that ever happened. And if that were true, it would really look rather grim for the future of the universe. For the kinds of reasons referenced above. If it suddenly were up to us alone to keep the show going on, we'd almost certainly screw it up.

Where the differing views are found, we see something a bit like a chicken-and-egg phenomenon. Some people think that we humans can and should construct an AI which then can take on a life of its own, become smarter than us, and guide our further evolution in a rational and benevolent way. I personally doubt that we're smart enough to do that right. But instead I believe that we already have such an AI, or rather an NI (Natural Intelligence), which has been active all along. Or many of them. And it is not so much a matter of inventing it, as of understanding it and accessing it.

I'd say that in the very big cosmological view, considering that time is just another dimension in a multi-dimensional universe, it ends up not mattering very much if it is one or the other. Is there a Bigger Intelligence that is hanging around from the beginning to ensure that things generally go in the right direction? Or do we invent such an entity and it exists from then on? Either way, it is there to find in the totality of the universe, and it is pretty inevitable that it will appear. Which to me means that, either way, it is an inherent property of the universe.

I am very skeptical of the view of evolution as only being random and blind selection. Natural selection is obviously part of it, but I'd rather lean in the direction of a larger definition of what is life. Autopoiesis, essentially. A galaxy or a planet can be "alive" in the sense that it maintains its boundaries and a certain equilibrium, even under changing conditions. Not all planets do that, but obviously some do.

To me, the expectation that humanity is likely to evolve a whole lot in a short period of time is based on the confidence that humans are part of bigger systems that are alive and which possess an inherent intelligence that is beyond us. Oh, the outcome might also be that humanity is decommissioned and a more useful intelligence emerges instead. Which might potentially be our computer AI. Or we might just go extinct and be replaced with some other natural species which is better suited. Either way, it appears that a higher level of manifest reflective intelligence is what is required next. Compared with animals who go about their business somewhat rationally, but who don't seem to have any abstract reflective thought about it.

What mostly seems to divide the people who think about these things into several camps is the fundamental assumptions about what consciousness is or isn't. I'll grant that intelligence is something that can be constructed with pattern matching and feedback loops of various kinds. But is it self-conscious? Some people say it doesn't matter, or that self-consciousness is just another pattern than can be constructed, or that self-conscious naturally emerges. I don't really believe that, and I think there's a factor that does make all the difference. Not that I can explain how that part (the (self-reflective) consciousness) is installed. I don't see any reason a machine shouldn't be able to have it, but I'm doubtful of the idea that it will just sort of happen if we just put together a complex program that does all the things we can observe intelligence doing.

Actually, I'd rather enjoy if we find out that there's a certain kind of equation that initiates consciousness. I just don't think we've been anywhere close to finding it. And I think that creating an AI without it would be a big mistake.

I think that human science at this point is suffering from a bit of a Sorceror's Apprentice syndrome. The hubris of thinking that because we've found a few magical formulas that can make things happen, we're big enough to be responsible for all the consequences. We can swap some DNA from one kind of cell out and pop in some DNA from some different cell, and after a large number of attempts like that, we come up with a carrot that insects don't like, and we think we've actually understood something very basic. Whereas we really still have no clue how to make life, or how it actually works, and the complexity is way beyond us at this point.

The same thing applies in many fields. We've found the traces of some wonderous things, and we've found that we can influence them, and we think we're just about ready to play God now.

The missing piece is the whole systems view. How a whole thing works in ways you can't fathom just from taking all the pieces apart and sorting them. And how it requires something quite more than just putting together all the ingredients to recreate and surpass some of the wonders that nature has excelled in for many billions of years.

It is like as if there's an ingredient X that is missing in almost all our technology. And that ingredient X is.... THE WHOLE THING, the full understanding of the systemic functioning and systemic effects of what we're creating.

We can mass-produce all sorts of wonderul technology, but we forget to figure out how to un-do much of it. We are terrible at creating sustainable technology. We can now create genetically engineered crops which have useful properties, but, oops, they got away and are creating havoc in the wild, because we didn't really understand how the natural cycles work. We can create nuclear bombs, but we forgot to invent the anti-dote, and, oops, suddenly some terrorist group has one of them. We follow the same stupid pattern over and over again. We invent a way of doing something, and it seems really useful at first, and we're really proud of ourselves, but we forgot to think through how it would really work within the eco-system of the world we live in, and we're incapable of cleaning up the mess.

What we've been excelling in so far are just smaller versions of the grey goo problem. So small that nature has mostly been able to keep up with the mess. But we're getting better at inventing more powerful, but still mindless and unsustainable technology. So it will not be for much longer than we can count on that we can get away with it relatively unscathed.

So there's a problem, a big problem. Some smart people think the answer to that problem is Creating a Friendly Artificial Intelligence. Making a super-powerful artificial intelligence program that by its design can't be anything but benevolent, and which, by using its superior intelligence and faultless logic, will keep us from destroying ourselves.

The trouble I see with that is that the people trying to construct it are using the same reductionist materialist thinking to create it that is bringing us deeper and deeper into an uncontrollable mess. And I'm afraid that what they're creating will be missing Ingredient X.

I have much greater confidence in Natural Intelligence. Not that I necessarily think that Nature will automatically and magically sort everything out for us. The playing fields are merging. Our human inventiveness is no longer separate from the rhytms of nature. Rather we seem to be becoming active evolutionary agents. What we are doing is in many ways the spearhead of evolution. Not a new better system to replace the slow random selection process. No, we are what nature has evolved. Now, what nature is evolving is to a large degree what we are evolving and developing and inventing. We have the option of being the conscious wavefront of evolution. Not just human evolution, but the evolution of life and matter on a much wider scale. We have the option, but no guarantee of success, if we misuse our newfound mental faculties.

A surefire way of screwing it all up is to ignore the wonderous brilliance of the plot so far, and arrogantly assume that whatever happened before was just stupid, random, blind stuff, and now we, as the first, will be doing something really clever. That's when the picture turns, and we come out as the ones who turn out stupid, random, blind technology. Carelessly letting cats out of a million bags that we have no idea how to get back. Just because we thought we emulated and greatly improved upon the half-assed workings of nature, when we didn't really get it at all.

Technology and artificial intelligence might very well be unavoidable components of our future evolution. But we have some catching up to do. We seem to be several orders of magnitude off from understanding the sustainable self-evolving design principles of nature. Principles that have brought forward the very consciousness and intelligence we're mis-using today. If we don't succeed in conjuring up the necessary wonder and humility and curiosity to discover how that actually came about, and how that already works, and how the WHOLE thing works - it is curtains for us, and probably for all life on this planet.

If we re-discover and re-connect with the inherent intelligence of whole systems, and get over our own pitiful self-importance, and we apply a much higher level of wisdom to our technological endeavors, then and only then might we succeed in manifesting the magical and amazing wonders that we suspect are just around the corner, and live to actually enjoy them, ... forever.

Posted by Flemming at April 3, 2004 11:35 AM | TrackBack