Does anyone here have an opinion about video game playing and its effects upon the development of world view and the self as participant? Maybe I am connecting dots that don't exist, but I am grasping for straws here.

There is such an ironic predominance of so called "Friendly AI" proponents in this group. Ironic because the same people seem obsessed with and actively building AI. Why build something if you want to kill it once it is built?

I ask this question about the effects of video game playing because I can't help but sense a personality profile that is some sort of strange amalgamation of paranoia and super-hero complex. How can you be a dragon slayer if there aren't any dragons? Well, you just build some dragons, that's how!

I doubt if this is the first time people have invented an enemy just so they could go to war. But in this case, the act is more obvious, blatant, and transparent than you'd find in a fairy tale or parable.

I had an interesting conversation after the last meetup. Was talking with a friendly AI freak who was saying that climate change didn't present a threat to the human species. I asked him what he thought life would be like for the humans remaining after the sea level had risen a foot or two and submerged most of the agricultural land and population centers? I asked him if he knew how to make shoes or grow food or build a water delivery system or run or maintain a power plant or dress out a deer or alchemize a batch of antibiotics or make a tooth brush or keep a woman alive during childbirth and he said it wouldn't be a problem because he could get that information from the net!

From the net? What net? I can't imagine a system more fragile and more dependent upon the smooth and constant operation of our most complex infrastructural systems (power production, power distribution, server maintenance, cooling, communication systems, satellite navigation, etc, etc, etc). Each winter, simple annual storms down a significant percentage of these systems and take weeks of very technical and energy intensive work by highly trained technicians with the exact right equipment and transportation to bring up to full use again. How long would a complex and fragile system like the web remain active or useful without the huge energy and human infrastructure that builds and maintains it?

Huge multinational energy, communications, and transportation companies routinely go under when demand or costs change by 10 percent or less. Looks like a mid sized power plant has between 100 and 200 employees. I doubt very much that any of these people are unnecessary. Companies have a tendency to keep their expenses as low as possible. Add to this all of the parts and supplies that must arrive on time every time to keep a power plant operational. Now add in all of the organizations that come together to make sure that power distribution systems remain operational. If any of these critical links fails, the whole system fails. And that is just power. All of our critical systems are inter-dependent. Food supply is completely dependent upon roads, fuel, trucking, trains, air transport, chemical production, irrigation systems and continent-wide water collection and distribution systems.

In the united states alone there are more than 100 thousand small farms that use industrial sized diesel engines to pull water from wells for irrigation. Big farms also use complex irrigation systems. And that is just farming. Our grocery stores would be empty in just 48 hours if they were not re-stocked every evening from a fleet of distribution trucks loaded up at regional distribution hubs. Our food is packed in paper, plastic, aluminum, and tin containers built every day at factories around the world, without which food would mostly parish on its way to the consumer. Everything is interconnected and interdependent. Knowledge is only a tailing in the overall complexity of which we move forward. Having a book on how to build a power plant brings you no close-er to your goal without the entirety of the infrastructure that allowed for the production of that book and its encoded knowledge.

I am having a lot of trouble imagining any situation in which AI could be built at all on a planet in the grips of climate change as predicted by the vast majority of those studying it. So maybe the best defense against non-friendly AI is climate change itself. Keep fueling up!

It is arguable that intelligence and the structures necessary to maintain it are the pinnacle of complexity in this (corner of the) universe. Certainly the infrastructural resources and stability that are necessary to understand intelligence as mechanism and process supersedes any project yet tackled by human kind. The idea that AI could be built in a decaying or decrepit culture with significantly less stability and resources seems absurd in the least. Maybe just maybe, the un-friendly AI dragons are already here… flying around and burning up reason and connection space within the confines of minds made susceptible by chronic exposure to fantasy and roll playing?

Climate change isn't so sexy a dragon as is fire breathing AI. Neither it seems are any of the other complexity-killers that loom so large in Earth's not to distant future – energy scarcity and cost, social and cultural instability, access to fresh water and air-able land, the rapid buildup of bio-reactive toxins, increasing population of adults from childhoods of extreme instability and abuse, four fifths of the planet gaining access to the power of science without an understanding of science, etc.

Dreams of the development of AI, benevolent or evil, teeter at the edge of the looming abyss that is cultural and resource chaos. If you care about complexity, than you have to honor the complexity that must be built and maintained as stable base before you can build higher. So often, when I witness AI related discussions, I am made aware of how rare it is to find complexity builders who understand or even acknowledge the need for an understanding of complexity.

Earth isn't a video game. Some reality please.

Randall Reetz

P.S. There is no way to control things smarter than you, so every mote of energy you spend in this futile direction is meaningless. We can't even control things (viruses, the weather, gophers on a golf corse) much more simple than ourselves. You want to be a super hero? Put your energy towards systems stability. Learn to measure the relative causal influence, the hierarchy of influence.

Randall,
I am in complete agreement with you regarding the foundational flaws of which so much is dependent. The success of future generations will depend on our present ability to foresee and react to situations such as those you describe. I see this discussion as an attempt at a little cold water in the face. As a group, mankind has not evolved much beyond the seeds that have allowed the very few of us to demonstrate what we consider to be intelligence. Time did not begin with man. The fact that the universe is not overcrowded with creatures that long ago surpassed our minimal understandings of what is intelligent, is testament to just how difficult it must be to proceed up the I.Q. ladder. Often it is not the brightest of us but those with the loudest egos that rise to positions of authority. And the vast majority of the rest seems to have taken some sort of social opiate and because of numbness, are only semiconscious. It is my hope that discussions such as this and the understanding offered as a result of the research regarding such disciplines as AI will offer some of the stimulus necessary for the wake-up call. But it is also our equal responsibility not to become “Chicken Littles” ourselves in the process. How are we going to continue this impossible life style and achieve these goals? What our free enterprise system has created as a definition of success is dependent on devastation to achieve its goals. Look no further than the kind of political ads that represent the candidates. In the successful ads, there is no discussion of opinion, only example after example of negative attacks and obvious lies. It’s what works for them and it’s a telling statement about us.

Lex Ricketts

Randall,
“There is no way to control things smarter than you, so every mote of energy you spend in this futile direction is meaningless.”

A couple of things about this. This would imply that the environment has no effect regarding animal behavior. I don’t think that you believe this to be true. Your definitions of evolution precludes this. In the animal world survival is allowed through environmental agreement. We have talked about motivation in relation to the animal world in the past and how it has a connection to survival. The obvious difference is that instead of nature describing the limits of the environment, in the machine world humans will in some form need to accomplish this. In this regard I don’t believe machine intelligence will develop without understanding the elements that allow it. I don’t also understand the mechanisms that lead you to believe that intelligence is possible without understanding and therefore controlling this. Or is what you are suggesting something like the plot of the 1970 movie “Colossus. The Forbin Project”. A super computer is designed to provide absolute world security for mankind. It is such a computational advancement that it develops self-actualized intelligence. It than sees mankind as a threat to mankind’s own security and enslaves it for its own protection. Beyond the obvious negative connotation there is another story about us. The problem with the enslavement outcome we unknowingly imbedded within primary objective of Colossus wasn’t with the computer. The underlying message here is that if an intelligence greater than our own were to evaluate mankind it would not like what it sees. It’s the same message about what God would do if he were alive. Presumably, he would punish the crap out of us. This is at best a narrow response. The greater intelligence sited in examples like these really aren’t that much greater if the limits of the entities responses are that narrow.

Lex Ricketts

Ok, there is a big difference between effect and control. We will effect AI. Of course. But control? Forget about it.

My friend went with her mother on an african safari. They were in land rovers observing a pride of lions. Big healthy lions. Lots of them. The jeeps had no doors and no tops. The lions layed there utterly indifferent to the jabbering humans and their cameras in their big metal contraptions just meters away. All of a sudden each of the lions looked off to the horizon to the north, shivering their backs. My friend asked one of the guides "What would scare them so?" The guide said, "See where that road disappears into the distance, soon you will see a dot that will become a tall very thin man in colorful clothing walking towards us with a stick in his hand." And sure enough in about 20 min. through the lenses of their binoculars, they were able to resolve this solitary human, walking bare foot and holding a long thin staff. The lions waited until he got within a mile or so and then trotted off quickly into the trees. She asked "Does he have a gun?" and the guide said, "No, he is a Masai herdsman." "As boys, their right of passage, is to go alone into the trees with a spear and kill a lion ."

This I believe is a great example of the difference between effect and control. The lions "effect" the lives of every thing made of meat that lives in the savanna. The Masai, thinner, smaller, weaker, slower, less protected, "control" the lions with a single pointy stick. Sure, the Masai are effected by the presence of lions, some are even killed by them, certainly their cattle are taken by the lions. But big brains allow the Masai to out-think the lion and this is the difference that makes the difference when comparing effect and control. In a very real sense, the Masai herd lion in the same way they herd cattle. By controlling the lions (without getting rid of them), they keep the lions around which is a very very very essential thing to do in a land also occupied by hyenas and other predators of less friendly dispositions.

All of this control is based upon the seemingly impossible action taken by a 120 pound adolescent boy. How does a young human overpower an animal 4 or 5 times his size and many many times his strength?

So I ask you, what will we build to mimic four inch teeth, massive jaws, forty, four inch claws on the ends of massively powerful limbs that can propel 500 pounds 50 miles per hour? And I tell you, it won't matter a wit. Yes we will be able to harass and even terminate some of these AI entities, but over all, in the long run, we will not control them. Never.

The better question is why we would want to? Why we think we should need to? Why we pre-imbue these things with evil or destructive or competitive intent.

There are other tribes that compete with the Masai in the cattle herding business. But their cultural habits don't work as well to control the dangers that lurk all around them. How do the lions know when the Masai are approaching? The Masai ritually cover themselves and their clothing with cow shit… they stink and stink travels farther and wider than any photon could!

How will AI choose to warn us not to come too close to their superior power? Human stink, pattern, sound, heat, long sticks? At the very least Lex, you are going to have to switch human and machine when you talk to the difference between object and environment. Superior intelligence will always lead to control. Think of a human and a tree. Which one is environment, and which is object? Why? They are both made of atoms, both built by DNA and both use oxygen and carbon exert structure. The tree is much bigger than the human, converts far more energy to structure and metabolism. Why is the human the object to the tree's environment? Because the human can out predict the tree. It is that simple. Next to AI, humans will be environment. Get used to it.

Prediction is equivalent to scope of input. The tree can hardly know the presence of the human, certainly not in the way the human can know the presence of the tree. Why is this true? Because the human can model the tree in a way that the tree can not model the human. The Masai boy can model the thinking of the lion, the lion learns (evolutionarily) to know and respect this imbalance. The lions of the Masai region are lions who are genetically predisposed to react in deference to this difference in modeling capacity.

It is funny to me that we have trouble imagining that some thing could know us better than we know us. But this is absolutely guaranteed. Every entity exhibits behavior that falls into pattern. Pattern is always more efficient. Without pattern we would consume infinite energy. We don't have infinite energy to consume. The more complex the control structure of the entity, the more complex these patterns. But patterns they remain. And patterns can be captured and modeled. This is true no matter how complex the causal mechanism from which they arise. This is why the behaviorists have the upper hand, they could care less what causes pattern, only what that pattern is. Causal understanding is something one needs in order to build, but it isn't necessary when modeling pattern. Anything that can model our pattern will be able to control us. Anything smarter than us will be able to capture our behavioral patterns. The only defense against such capacity is novelty and novelty is too expensive, we can do it, for a while, but eventually we will fall back into pattern, if only to recoup our energy.

Randall

So, what could we do with our energy and effort if we didn't have to spend that energy and effort on paranoia and defensive planning?

What if we just accept the fact that we will not always exist, that nothing exists forever, that we are here to bring the next thing into existence. What if we just do our job as everything before us did its job?

Isn't that life enough? Isn't that life?

As for your observation that "loud egos" sometimes win in battles with reason and truth… this is only true in the short term, and locally. In the long run, in the wide scope, truth always wins. Truth is less expensive to maintain. Truth is more stable. Things structured closer to the deepest causal truths eat up less energy as they build complexity. Which ultimately allows them to survive long enough and become pervasive enough to enable bigger and more far reaching and faster use of energy. Don't sweat the egos. Ego plays a dangerous and self-destructive game. Said another way, truth does not depend on ego.

Randall

Randall,
“and this is the difference that makes the difference when comparing effect and control”

I don’t see this as a comparison of effect and control. However, I would agree with your perspective regarding this if the degree of intelligence remained at this level. The relationship of Masai and the lions revolves around the meaning of the lions to the Masai. The invention of lions weren’t totally due to the understanding that the Masai had of them. This is a totally different level of control. But more importantly is the reflection of our own motivation that threatens us. It is this we so readily recognize as a potential threat. Are the contents of our concerns based on what we would be doing if we possessed this greater intelligence and this kind of control was within our grasp. The concerns of an intelligent machine will be a product of what motivates it and this will be well within our control. In the words directly from a machine:

“More than iron, more than lead, more than gold I need electricity.
I need it more than I need lamb or pork or lettuce or cucumber.
I need it for my dreams”
from The Policeman’s Beard is Half Constructed By Rader

This says it all to me! “I need it for my dreams” and you would argue against this. How dare you!

Lex Ricketts

Lex,

I am concerned by the naiveté exposed in your statement:

"The concerns of an intelligent machine will be a product of what motivates it and this will be well within our control."

Some naiveté is cute, but come on man! First, there is the math (Godel) proving that nothing can ever come close to understanding something as complex as itself (let alone building it). So whichever intelligent "machine" you are talking about, won't be of our "design". If AI is ever to breath its first breath, it will have to have self-evolved from a very simple starting point. By the time it gets "intelligent" we humans will have absolutely no idea what it is made of or how it works, or frankly, any profound ability to ever know. Reality.

I am sick of this very awkward form of "Rock Star" fueled nerd machismo. It won't matter if AI is friendly or unfriendly, at least this time, the nerds are not going to save the world. We are going to have to find some other way to deal with the stuffed anger and deep sense of humiliation high school left us with. Video game anyone?

Pry a little on the schism that is made obvious by this discussion (and its endless recursion) and you expose the vast chasm between technical know-how in the rarified and extremely abstract world of code-writing, and knowledge of the sort that gives cohesive perspective to these larger-than-software concepts. To properly frame AI one must have a firm grip on big knowledge; knowledge of the biggest most foundational rules by which this Universe works. Software is structured by rules that do not map well to the much deeper level that is causality. You can derive software definitively from physics, but any software derivation of physics (simulation) will stand beside an infinite number of other derivations who's veracity can only be checked through constant measurement informed calibration. The rules that govern abstraction systems are far more flexible and far less causal than the real systems they abstract. That is a good thing for software writers, but a really bad thing if those software writers can't see where their precious and protected world of simplified logic ends and where causal foundational reality begins. Hubris at its very most self-blindingly obvious worst.

Early computer scientists, scientists who didn't call themselves computer scientists, scientists that got into computing as an extension of logic or physics or mathematics or topology, had a strong pre-existing epistemological or cognitive knowledge map upon which was layered what they learned while inventing artificial computation and its most basic concepts. Today's engineers often start writing software at an early age, and only later, reluctantly, to fill out the university requirements, learn physics or cosmology or evolution or thermodynamics or mathematics or logic. That type of personal history results in a different brain, a brain with abstraction at base and causality strapped on top like luggage.

Randall

Randall,

Regarding your opening statement on dragon slaying, I don’t have an impression that those in the group who are concerned about the “friendly AI” issue are simply looking for dragons to slay. I think they’re concerned about what they see as a genuine, important issue. But I do think that there are a lot of misconceptions and bad assumptions being made (on all sides) as to what AGI would look like and how it may come about.

Since we have no theory of AGI and don’t know how it arises, we reason by analogy. We see ourselves as examples of intelligent agents, and so conceptualize AGIs as vaguely “like us, but smarter”. We expect any differences to be related to their being “smarter”, and therefore (as we see it) having goals and motivations that are incomprehensible. To me, that’s complete bunk!

General intelligence – whatever it may be – has nothing directly to do with goals and motivations. It may give an ability to predict the consequences of actions, but it doesn’t determine the goals. Not at the top level. Given a set of goals, it may derive a set of logical sub-goals it can act upon to advance the given goals, but the top level goals must come first.

Most of us in this meetup, I think, agree with Monica that AI's worthy of the name will not be neatly designed in the way we currently do most computer programs. They will need to be evolved. But a point to be clear about is that evolution does not just happen by itself; it requires selection. Selection drives it toward fitness under whatever selcetion criteria are operational. For nature, the selection criteria boil down to just one thing: reproductive success. Specifically, long term reproductive success of heritable traits in a complex environment of interacting trait carriers.

When we employ artificial evolution in a simulated environment to develop more effective programs, we employ selection criteria that are a lot more specific and goal-oriented than nature’s. If we’re interested in parsing the symbols on Chinese web pages, we’ll write a function that evaluates candidates on how well they do that job, and we’ll use that to select which candidates to retain for breeding the next generation of candidates. If we’re lucky, what will emerge from the process will be what we wanted – a program that’s good at parsing Chinese web pages. We may or may not be able to see how it does it, but we can be certain we won’t get anything that we didn’t select for. That would be like creating a pattern of random light and dark regions on a film, and expecting the result to be a hologram! (A hologram, when viewed at high magnification, does in fact look like a random pattern of light and dark regions. The connections in a trained neural network will also look pretty random. But neither holograms nor trained neural nets are random, however much they may look like it to us. And the probability of either arising spontaneously is essentially zero.)

The bottom line is that the selection rules and evaluation functions that direct artificial evolution are just as much design constraints as any rules we follow when writing a program. Any AGI, whether programmed or evolved, can't help but be a "designed entity". It will have the goals and motivations that were designed into it, either by conventional programming or by programming of the selection criteria.

There is zero chance that hostile AGI's might evolve by accident. If they ever do come into existence, it will be because somebody with a grudge against humanity went to a lot of work to create them. That could certainly happen, but it's a social problem, akin to the "terrorist with a nuclear bomb" scenario, rather than a technical problem.

Randall,
I remember back in 75 or 76 when I first came up with the core of my present thinking. It was well accepted that in order for a machine to replicate human thought it would require the program representation of all responses that a human would ever be able to demonstrate. I’m sure you’re familiar with this. At that time my ideas centered around computer learning with the focus being on how psychology could be used to provide direction. To me this was an obvious way to avert the Herculean task of providing an all encompassing program. My discussions at that time were very much similar to the ones I am having today, with one exception. Computer learning has caught on. This is what confuses me regarding your comments about my naiveté. Do you think that my naiveté has improved. Back then I was accused of being simplistic and naïve for using computer and learning in the same sentence. So with the acceptance of computer learning I must be improving. I’m hoping that if I hang there in a few more years I won’t be naïve at all. One can only hope.

Man, you had to bring up Godel! Damn, now that’s some complex thinkin there! Lucky for me you pointed that out. So I guess we can forget about humans. But what about monkeys? Can we understand them? I’d settle for that. Or maybe babies or even better yet, baby monkeys? They should be less complex. Would the math work then? Or is this line of reason simply pointless and only serves to avoid the issues.

“By the time it gets "intelligent" we humans will have absolutely no idea what it is made of or how it works, or frankly, any profound ability to ever know.”

Randall I have to ask the question. Is this statement about some future AI or is it simply a statement about you, here and now. Because to me it sounds like you might not have a clue about anything regarding the development of intelligence, nor does it seem you will be open to getting one in the near further. You don’t mind if I proceed however. I don’t think it will reflect poorly on you or anything.

“To properly frame AI one must have a firm grip on big knowledge; knowledge of the biggest most foundational rules by which this Universe works.”

Conclusions such as this and most of the others you attempt to develop here seem to presuppose the reality of what is in fact. What is in fact about AI is that there is nothing in fact about AI and these are some very broad assumptions you make. Have you been in contact with people from the future? Because these statements you make seem to be very factual and almost clairvoyant. This is what I imagine the situation that surrounded the 1500s discussion about the earth being flat would have been like. The vast majority of the intelligentsia of the time was convinced that the world was flat. These were bright individuals and I would imagine they had very convincing arguments. I would like to have witnessed how convincing reasoning faired in contrast to the power of the socially convenient argument of a flat earth. From my perspective not very well.

You refer to my arguments as having “endless recursion”. You’re making an excellent point here. I’m concerned about that myself. This group does not seem to share enough of a meaningful background with me. This doesn’t allow us to share common references. References necessary for understanding. I have attempted to build these references but I seem to lack meaningful technique.

Lex,

There is a difference between a complex but handleable project, lets say building a manned moon-rocket, and a project that requires a level of complexity that outstrips human capacity. I have my own theories which I like to think will be true, at least in gross measure, of any intelligence, but these are not a lot different from an "artificial" flight enthusiast in the 1800's exclaiming: "Any device that might someday fly will have to produce more lift than its own weight!" Though true, such a statement doesn't tell you how. And then there is the issue of the absolute lower limit of the complexity that must be assembled together for a system to become intelligent. I have trouble believing that evolution, having iterated on complexity for billions of years, exploring endless tangents, has settled onto in a form of intelligence that is grossly inefficient, or that we could gain great advantage over "nature" and create a thinking machine that is radically more efficient (needs less parts, or that these parts could be more simply associated).

There is an interesting story, involving the MIT fueled startup "Thinking Machines". They architected a hyper-parallel computer with thousands of tightly networked processing cells. Leveraging the authority MIT and their own distinguished careers lent them, they procured a contract before they had ever built a machine. With cash in hand they rented a warehouse, purchased the requisite chips and electronics, and set to work assembling their first computer. After a few weeks working as fast as they could, they realized that the process was much more labor intensive than previously envisioned. Multiplying the amount of time they hand currently spent times the percentage of the task left to be completed, they calculated that it would take them another decade to finnish this one computer. Back to the drawing board. This time, they designed the computer for efficient assembly. The second design allowed completion in less than two months.

Complexity is equivalent to effort required to assemble, which is equivalent to how much causality will resist the accumulation of a set of elements towards a particularly unlikely configuration. When faced with the task of building a complex system, it isn't so much an issue of understanding the full finished system… rather it is the ability to define a process by which that particular accumulation could come into being. As project complexity increases this becomes a larger and larger issue. Designing a moon rocket is one thing, building the nation-wide human resources and technical apparatus by which it will be accomplished is quite another.

If I am correct, it doesn't mean that, in some gross way, we won't ever be able to understand intelligence as a general topic. But intelligence like flight must be more than theory, it must be instantiated in physical space, it must be operational. Though E=mC^2 might perfectly model the relationship between mass, energy, time and distance, it would be ridiculous to claim that knowing E=mC^2 would allow a person to build a universe. Having a model for intelligence will likewise get us no closer to the ability to build it. The veracity of a formula is best understood the way that we accept that an equation can be created that fits an empirically collected data set. Obviously, statistical methods, no matter how perfectly applied, are not the same as the causal effects that worked together on time and matter to produce that which can then be measured as data. Even a perfect abstraction of a causal system, one that always predicts reality, isn't the same the causal reality it maps. Einstein's equations don't make a galaxy… galaxies make Einstein's equations.

So, I am not saying that we will not some day have some generalists understanding of what intelligence is (or more likely, what it isn't), I am saying that building it is never going to be a wrote process in the way, say, the Apollo rocket was.

This is because intelligence is at least as complex as the machine (our brains) we will be trying to build it from. The Apollo rocket wasn't nearly as complex as the brains from which it was understood and constructed.

I have spent my life working towards an understanding of intelligence and complexity. I think I have acquired a good base of knowledge, a good scaffolding from which to stand in order to better see the parameters and limits involved.

In the process I have begun to see the difference between hard and impossible. I certainly believe that it is unlikely that human intelligence defines a limit to the complexity achievable in this universe. As such, I look into the future and see an advance in the rate of the evolution of complexity handling that looks very similar to the progress made to this point. Will "artificial" intelligence come into existence? Guaranteed. Will it come into existence because of the effort of humans? Of course! Does that mean that we will need to understand intelligence in order to design intelligence? Of course not. Do you need to understand plants to grow a garden full of healthy plants? No you don't.

Do you believe that humans have between 100 billion and 100000 billion brain cells but only need some small fraction of those cells in order to think? Nature can't afford waste.

While you are trying to "architect" intelligence, I will instead be attacking the problem as a gardener would. I will be asking the bigger questions, ignoring as arrogant our natural attraction to details. The only way to build something more complex than ourselves is evolution. If you look back in time, you have to believe that this "evolutionary" method works in all situations in which that which is built is more complex than that from which it is built. Evolution comes with an impressive resume. There is proof that it works to build space time from ?, to build matter from energy, stars from plasma, to build heavy elements from stars, to build life from chemistry, to build intelligence from life.

So, if you believe that intelligence is less complex than the intelligence (your own brain) from which you are building it… than by all means proceed in the roll of "architect" or "designer".

Of course I wear the same hats, but the design and architectural tools I choose to wield are those of the evolutionary process. Obviously, intelligence isn't a pre-requisite of intelligence. You don't need a fish in order to evolve one. Else we would never have been here. This process I believe to be the real pattern of intelligence.

I ask, what is it about the evolution of galaxies that is self-similar to the evolution of concepts within an abstraction machine (brain)? And I believe the answer to this question provides a more accurate understanding of intelligence than will any study of abstraction machines on their own.

Randall Reetz