Supermodelling

It seems to me that Mind and the Universe are tightly connected (beyond the obvious point that Mind is part of the Universe!):  minds predict the universe, manipulate it to achieve goals, reason about the universe in ways that are not quite the same as “prediction”, and so on.  Summed together these roughly cover what we mean by “understanding”.

Another way to put this is that mind models aspects of the universe, and that is the way I normally think about it.  This rather trivial starting point is already extremely vague.  I am not trying to define mind, model, reasoning, and so on with any sort of rigor that could lead to a formalization.  For now just know that I consider such formal methods to be rather limited in their usefulness (although exteremely powerful when applicable) — and I will explain why in due course.

If a mind models the universe and we want to build a mind, it’s useful to know what the universe is like.  Without a viewpoint on that issue, it is difficult to invent appropriate modelling techniques.

There are many possible ways of looking at such a broad question as “What is the nature of the universe?”  The most obvious (to a guy like me living in my culture) is to say that the universe is a bunch of “atomic” entities that behave according to concise mathematical rules.  This physical reductionist program has been very effective for scientists.

That’s not the way I want to attack the issue, though — at least not as a starting point.  I’ll save discussion on why until later, but basically this approach is inapplicable to many things that minds care about, and intractable for most things to which it could theoretically be applied.

Instead, I’d like to start with the really obvious points.  The blazingly obvious interests and amuses me because I do not really understand very much.

What I mean by universe includes more than physical objects and processes (for example, it includes mathematics).  However, as a starting point I’ll limit myself to the physical aspects of the universe.

So:  There is stuff in the universe, and that stuff is often non-uniform.  The non-uniformities are in fact quite peculiar.  Most astonishingly, localized globs of stuff often have radically different properties.

Definition:  A localized glob of universe-stuff with statistically highly nonuniform (with respect to surrounding stuff), non-random properties will be called a thing.

I suppose this must seem like the most brain-dead sentence ever written, but I find it astonishing that the universe has things in it!  I’m completely blown away by this fact.  When I go out running and look around at the scenery, the thinginess of the universe just amazes me.

I don’t want to attack the issue of why the universe has things in it.  There are many fascinating aspects to that question (the Anthropic Principle for example).  For my purposes, the universe is accepted as is.

Here are some things:

Having focused on things, I am immediately tempted to come up with a formal definition of thinginess, but again this does not serve my goals here.  What I am really interested in is the relationship between minds and things — they way things may be apprehended by minds.  Before I can come up with an approach for that project, though, I need to think about more types of things.

A casual interest in electronics and mechanical devices led me to play around with robotics as a hobby for a number of years.  Much of that time was spent on what we generously called “combat robots”, which are actually radio-controlled vehicles which inflict senseless and amusing carnage on each other inside lexan-clad arenas.  I was also interested in walking robots and spent a lot of effort playing around with machines with many degrees of freedom.  I built many such machines and learned a lot about electromechanical devices.  For the interested, I described a lot of that stuff here.

My AGI viewpoint holds that intelligence should to a large extent self-organize through adaptive interaction with a rich and causally deep world (I will have much more to say about that in future posts).  That is, most of the complexity of an intelligent entity is most efficiently acquired from an environment rather than being pre-specified.

Since the actual world is a pretty handy environment that has the Right Stuff built into it, it’s natural to think that embodying an AGI project in a robot with capable sensors and effectors is a good approach and I have been intending to do exactly that.  I’ve been collecting fancy parts like servo motors, harmonic drive gearboxes, and high tech motor controllers from ebay for several years, I have built a computer controlled milling machine to help with the construction, and I have enough parts to build a robust quadrupedal research platform totaling 14 or 15 degrees of freedom, with cameras, touch sensors, and so on.

Few researchers explicitly studying AGI actually use robots as their testbeds as far as I know.  However, there are many “AGI-related” research efforts that use robots.  Some examples:

• A large multidisciplinary European research track called RobotCub is one such program, based around a public high-degree-of-freedom humanoid robot design.
• Juyang Weng’s SAIL projects have a definite proto-AGI flavor to them.
• MIT’s Cog project is an ambitious effort to study various aspects of embodied human-type cognition.
• Bob Mottram is a roboticist who is also a frequent contributor to the AGI community.
• Many other mainstream robotic efforts have natural connections to AGI.

One thing makes me think that it might be a good idea to wait a couple more years before building an ambitious robot, though:

Although video cameras provide a nice dense connection to the world, it may be that the lowest-level learned concepts are much easier to acquire when similarly-dense haptic and proprioceptive sensors are added to the mix.  These not only provide an immedite and detailed interface to a complex world, but also provide rich cross-modal integration with vision.  It seems to me now that rooting out the secrets of early concept acquisition would benefit greatly from that sensory modality.

Problem is, robots suck at having a senses of touch.  For example, artificial skin is poorly developed and out of the technological range of an amateur like myself.

There is an alternative that is much in the minds of AGI researchers these days:  virtual embodiment in simulated worlds.  One of my biggest research topics these days is evaluating the potential of that option with respect to the needs of my own exploratory research.

I shall have much to say about that as my investigations proceed.

Hi!  I became intereseted in Artificial Intelligence in the late 1980’s when, after several years on academic staff at a large midwestern university, I entered the graduate program there.

I wanted to help build a thinking machine.  A real artificial intelligence, the stuff of scientific fantasy.  The years I spent thinking about this were very interesting, but they did not result in a doctorate.  Unwilling to work on “small” problems, I sketched out what I thought was a reasonable research program for eventually producing “real AI”, but I wasn’t talented and smart enough to produce any impressive results.  Although my thinking has drifted a great deal in the 15 years that have passed since I left academia, the basic outline in my head is still more or less intact.  I’ll describe it in pieces over time in this blog.

Since then, I’ve had a career in “industry”, building things with economic value in exchange for money and appreciation.  It’s a pretty good gig.  Although I occasionally have taken out the old notebooks, done some reading on current AI research, and in other ways kept the dream alive, mostly I have been watching computing power grow.  I believe we are still maybe 10 years away from truly interesting “baby” general AI systems (and further still from systems powerful enough to have a large impact on society).  Still, barring catastrophe, I should live to see it, and I think it’s time to pay attention again, and participate in my own modest way.

A field named “Artificial General Intelligence” has appeared, with some amazingly bright and insightful thinkers working hard (along with the expected crackpots).  Even though I am only an amateur, I would like to participate in this field.  So far my participation has been limited to some posts on mailing lists.

With the rise of the Web, access to intellectual activity at all levels of rigor (from papers in prestigous journals all the way to mailing list rants) is easier than ever before.  One thing I get particular pleasure from is blog essays, in which individuals lay out their world view for others to take whatever they find useful in building their own castles of thought.

So, although I have not “solved AI”, it will be good for me to write down my thoughts — for my own benefit, and in hopes that somebody out there in cyberspace will find something of interest.

Some of what I write will involve slowly explaining and refining my own still admittedly rather vague point of view, and some will be in reaction to the work of others.  And some will be off-topic (it’s my blog and I’ll write what I want!)

I hope it ends up being useful in some way.