Tuesday, February 24, 2009

Are you human?

I was thinking about capchas. You know those sequences of letters that you have to enter in order to post a comment to this blog, for example. And I was thinking what a sad statement it is from an AI perspective that computers aren't able to read these.

Then I remembered a comment I came across somewhere next to a capcha - "Are you human?". You see that is exactly the purpose of a capcha - to distinguish between a human and a machine. It is assumed that a machine, in the form of a javascript, will be incapable of reading the capcha. And this will stop someone from writing a script to automatically go out and write spam comments or buy up all the tickets to the latest Bruce Springsteen concert.

But what this basically says is that current OCR (Optical Character Recognition) software is incapable of reading some slightly mangled characters with a noisy background. How sad. I remember buying my first flat bed scanner that came with OCR software and thinking that I would easily be able to scan in a page of text from a book and have the OCR software read it in. It turned out the OCR software was about 80 or 90 percent accurate. But manually fixing the last 10 percent was almost as much trouble as typing in the whole thing from scratch.

This was perhaps 15 years ago and I assume that OCR technology has come a long way since then. But apparently not far enough to read a simple capcha. Because if you think about it, all that is necessary is to have a script that grabs the capcha image; runs it through an OCR program to get the text; and then inserts the text in a field on the webpage. Sounds pretty simple. Except that the OCR software is apparently not up to the task.

OK, but eventually there will be OCR software that is quite capable of performing this task that will run on a home PC. Then what? Well we won't be able to use a capcha to distinguish between a human and a machine. So some clever person will have to come up with a different way. But if you take this process to its logical extreme you end up with a Turing test. So at that point can you say that you have an AI?

I've always thought that the concept of the Turing test is flawed. The concept is absurdly vague and unscientific. I wonder whether Turing ever expected anyone to take him seriously. The whole thing seems like a joke that people have accepted as if it were some profound Truth.

There are parts to being human which have nothing to do with intelligence. An AI could be perfectly intelligent without having a physical body that resembles a human and therefore be completely unaware of many areas of human experience that would have no relevance to its existence. Should we then come to the conclusion that this AI is not intelligent?

My suspicion is that the first true AI will exist in some sort of cyber-space rather than in the form of a robot. That AI could exist in some sort of simulated world. Or it could exist as an intelligent agent (IA) on the web. Either way, it would not have a body at all.

Even if it were a robot, it could be some sort of industrial robot used in manufacturing. In that case it would probably not be humanoid at all. Such a robot could be quite intelligent without passing a Turing test. It might not even be able to speak in a human language. Would that disqualify it from being intelligent?

It seems that the Turing test is really a sort of "racist" test of intelligence. It tests whether the AI can duplicate the intelligence of a member of the human race, but clearly there are other types of intelligence. And certainly it is to be anticipated that an advanced AI could be far more intelligent than any human.

At that point humans would fail the AI's own version of the Turing test. From this advanced AI's perspective, humans would be considered biological machines of limited information processing capability - but certainly not intelligent. Perhaps much in the same way that we look upon other animals. You wouldn't consider a dog or a monkey as intelligent - would you?

Thursday, February 19, 2009

Capitalism and Progress

Here's my latest comment which I posted to Kevin Kelley's Technium website. I find that I have to work very hard to keep my comments from getting too long. That was the main reason I created this blog - so that I could elaborate on some of those comments.
Hi Kevin,

I’m sure you are aware, but perhaps you have forgotten, that Bill Joy in his classic warning of the dangers of a Singularity, “Why the future doesn’t need us” also mentions the Unabomber. I believe the reaction to his article was quite harsh and he was viewed as a sort of neo-Luddite.

Personally I’m open to criticisms of the “death march” of Progress. There are certainly benefits to mankind from technological progress, but the dangers should not be overlooked as they often are.

You mention the Amish in this article. I never quite got around to commenting on the full article you wrote about the Amish, but I think there is an analogy to the way that the Catholic Church maintained control over society in the Middle Ages. Just imagine if a young Galileo was born among the Amish. How would they handle the situation? Anyway, I think the last chance western society had to stop the unrelenting march of progress was Galileo. From there the combination of science and capitalism launched the revolution which has shaped the world we live in today. Like scientific progress, capitalism is built on an exponential model. These are dynamic systems that collapse under static restraints.

Are we willing to give up progress for a stable world system? I don’t think we need to return to a hunter-gatherer lifestyle, but it would require limits on technological advancement.

BTW, I’m so inspired by your writing in Technium that I’ve started a blog to record some of my reactions and responses to your articles.

One thing I've noticed about Kevin's articles is that he nicely keeps the discussion focused on the technical and cultural questions without somehow getting into the realm of politics. I'm nowhere near as diplomatic in these areas as Kevin is and I tend to drift easily into politics. Certainly if we are talking about the Singularity then the question of politics is eventually unavoidable. It will largely be up to the governments of the world to decide how that Singularity will emerge. For instance, the US military is already actively involved in AI research through DARPA.

But the issue I wanted to discuss here is the one I brought up in my comment regarding the connection between Capitalism and Scientific Progress. I mentioned that they both are built on an exponential model. I couldn't go in depth on that in the comment without making it way to long so I'll do that here.

First Capitalism. My view of Capitalism is that it requires ever expanding markets to survive. It is expected that the stock price of a company will continue to go up. For this to be true the company's net worth must continually grow. And therefore the company must seek out ever growing markets. If the company is not growing, then eventually it will go out of business - cease to exist.

An alternative is to reduce costs. Here is where Progress comes in. Through Scientific Progress a company may find a way to produce its products at reduced cost. This allows the company to continue to grow and avoid a virtual death in the business world. Therefore there is a marriage of convenience between business and technology.

John D. Rockefeller is the epitome of applying technology to business in order to reduce costs. As a result he became the richest man in history and created the world's most powerful company, Standard Oil.

It is actually quite surprising to me that today there is little emphasis by business on artificial intelligence. Yet I believe this is where the future of business is. I would have hoped that in the recently announced stimulus package there would have been a significant portion to developing AI. This is where America needs to be investing.

Even though Japan is currently in a very bad economic situation, I believe that it will come out ahead because of its attention to the area of robotics. Actually I believe this strategy is slightly flawed. I think robotics gets too much attention, whereas the real attention should be on AI.

AI does not require a physical body as in a robot. AI can have a virtual body and exist on the internet. So much of our world already exists in virtual space, that an AI living on the internet would hardly be hampered - the financial world is particularly accessible through cyberspace.

Next Scientific Progress. I should hardly need to say anything here. Moore's Law pretty much says it all - technology is progressing at an exponential rate. Just when we think that technology has reached its limits, knew scientific discoveries are made to maintain the rate of progress.

Certainly if we use biological organisms as our guide then there is a whole world of technology available that is capable of changing our world radically. The technology of life is nanotechnology. That is the secret of life. It manipulates materials at the molecular level. Nanotechnology is one of the game changing technologies that can eventually lead to a Singularity.

Bill Joy refers to these technologies as GNR - genetics, nanotechnology, and robotics. I would prefer AI to robotics, so perhaps a better acronym would be GAIN - genetics, artificial intelligence, and nanotechnology. Joy then talks about KMD - knowledge-enabled mass destruction. I prefer calling the term WMD-H - weapons of mass dehumanization.

Of the three, genetics has the head start. We have already begun tinkering with the existing blueprint of life - DNA. This seems to me a perilous road to take. I actually welcomed the decision by Bush to stop stem cell research. This is an example where government deliberately slowed the pace of technology. I don't think the reasoning was quite the right one. I think its dangerous to mix religion with politics. The reasoning should have been more along ethical and moral lines without resorting to a particular religious point of view. The ultimate concern should have been the impact that such a technology could have on society, and not so much whether it violated the Biblical interpretations of one Christian denomination or another.

Once AI and nanotechnology reach the state of maturity that genetics is at, then there will be similar moral questions. The mature combination of the three GAIN technologies will of course be the point of greatest danger.

Personally, for me, the key technology is AI. Once a true AI is developed, that will impact the other GAIN technologies. AI will be turned on genetics leading to significant improvements there. And AI will also enable nanotechnology to finally become viable. And it is a given that AI will also be used to develop more powerful AI. This is one of the ways that exponential growth continues in technology, by using today's technology to create tomorrow's technology.

There is no incentive on the part of Capitalism or Scientific Progress to slow down this juggernaut. Bill Joy has proposed that we begin to study the problem now and plan for the future, but his good advice seems to have gone unheeded. While the atomic age was thrust upon us overnight, there have been predictions of the coming of the new age of AI for many years now. Mankind has no excuse for not preparing for this eventual outcome.

See also Bill Joy's 2000 article in Wired magazine, “Why the future doesn’t need us”.

The "alien" technology of life

I finished reading the latest Technium entry about the Unabomber - of all things. As I was reading through the comments I came upon one comment that touched on the all too familiar question of what is natural - is a road that Man makes natural?

This in a very obtuse way triggered my mind to bring to consciousness a recurring personal idea which had laid dormant for a while. The idea is that life consists of "alien technology". This is certainly not a 100% original idea - but then what is? Think of the Arthur C. Clarke classic "2001: A Space Odyssey" for example.

I am not a proponent of the idea that life was imported from some alien civilization. If for no other reason than this doesn't at all solve the problem of the origin of life. That's not my point here at all. My point is that even today, we do not fully understand the technology that supports life. At the lowest and mechanical level we cannot create life from simple molecules. And at the highest and most abstract level we cannot create an artificial intelligence.

In this sense the technology of life is completely "alien" to us. If we view biological organisms as machines, we must admit that they are far beyond our current capability to fully comprehend - even the simplest one celled organism is amazingly complex.

Here is my scenario. Imagine a mechanical robot-like pseudo-human species that possesses our knowledge of technology except that it has never encountered earth-type biological organisms. Suddenly from the sky a spaceship descends and out comes an elephant. Assume for the moment that these pseudo-humans thought of this elephant as some sort of machine rather than a "living being". That elephant would represent a huge technological challenge for them to understand. It would be viewed by them as consisting of a technology that was totally "alien".

We are so conditioned to separate biology from the other sciences, that we under appreciate biology as a technology. I am not trying to trivialize the spiritual aspects of life by saying this. I find that the "miracle of life" when viewed from a purely technological viewpoint is no less "miraculous". It is every bit as awe inspiring and mysterious - but in a different way I suppose. To me the more I learn about biology, the more in awe I become.

These are systems that are far more complicated than the most complicated software systems that we have yet invented. The immense number of interactions that take place in a simple cell are astounding; to create a multi-cell organism is orders of magnitude more complex. Finally, the creation of organs and circulatory systems and nervous systems and endocrine systems absolutely boggles the mind. Add to this the ability to reproduce and I find myself reduced to a reverential state of wonder.

Will any human inspired technology ever be able to surpass the "alien" technology of life?

Sunday, February 8, 2009

Michelangelo's depiction of the brain

After I finished my first post titled "In the beginning", I was looking at it when in an epiphany I realized that the shape of the "cloud" that God is in looks like a human brain. At first I thought I was reading too much into this, but the more I looked at it the more I came to feel that this was some sort of artistic symbolism that Michelangelo had employed in his iconic fresco. I felt as if I had discovered something important, but still was not quite sure of the connection.

Then I was reading through an article on AI and came across a picture of a human brain and immediately I flashed back to the depiction of God and Adam. This time I felt the connection was so strong that it must be obvious and it must have occurred to someone before. So I did a quick Google search of "adam creation michelangelo god brain" and bingo! There it was. In 1990 Dr. Frank Lynn Meshberger had made the same exact observation in an article titled An interpretation of Michelangelo's Creation of Adam Based on Neuroanatomy". (JAMA 264:1837-1841, 1990)

[I was]... immediately struck by the shape of the image surrounding God and the angels. It was the same thing I had been working with all day! It was the unmistakable outline of the mid-sagittal cross-section of a human brain.

So my choice of image was much more significant than I could have possibly imagined. But what did Michelangelo intend it to mean? This we will never know. One interpretation is that it represents God as a higher intelligence which then bestows this intelligence on Man - "in his own image". It is after all intellect which distinguishes Man from the "Beasts".

Or it could be that Michelangelo is implying that God exists in the brain of Man. This might have been a popular philosophical undercurrent during the Renaissance, although it would seem to be more appropriate to the Age of Reason in the 17th century which followed the Renaissance. In Michelangelo's time, such a view would have been deemed heretical. Would Michelangelo have risked expressing such an unholy concept, even if done through symbolism?

The first clue for me was the dangling foot which seemed so appallingly unaesthetic. Meshberger associates this with the pituitary gland. The interweaving figures which support God seem to represent the folds in the outer gray matter of the brain. Meshberger with his knowledge of neuroanatomy goes into much greater detail.

Until I looked through the transparency I didn't realise that one of the angel's backs was the pons, that the legs and hips were the spinal cord... The knee of the flexed right leg of the angel with the bifid foot represents the transected optic chiasm, the thigh the optic nerve and the leg itself the optic tract...

If Michelangelo were alive today, perhaps he would make a movie depicting Man imparting life to his AI creation.

In the beginning

The purpose of this blog is to talk about artificial intelligence - especially as it relates to evolutionary algorithms. Therefore the name, the evolutionary brain or evobrain for short. I have also taken on the nom de plume of "evobrain" which might get confusing, but there it is. I suppose a true AI could easily figure out from the context which evobrain I was referring to in the great majority of the cases. Ambiguities, it seems to me, is one of those areas where the human brain is vastly superior to current computer logic.

The origins of this project was a program I started writing a few years ago about a mouse in a maze. The idea was that the mouse should be able to learn to traverse the maze more quickly and then be able to apply that learning to another maze. Eventually it would come upon a algorithm for traversing mazes.

I intended this to be a very simple example of artificial intelligence which I could easily simulate. The idea was that this would provide me a platform to run AI experiments and to develop my own ideas on the subject. I thought about working with some sort of real physical robot, but decided that simulation would be a much more efficient environment in which to test my ideas.

I settled on Java as a programming language because not only could I write the logic, but I could also create a graphical simulation with a minimal amount of effort. I already knew C and some C++ so it was not that difficult to transition to Java - although it's hardly a trivial exercise. I had never done GUI programming. It was something I had wanted to do for a long time, but I was never happy with the Microsoft approach. When I started looking at how Java incorporated GUI, it just made sense to me and I liked that I could run the same program on various platforms. This became especially important when I bought my first Apple computer which is the MacBook that I'm using to write this blog.

The evolutionary part of the name "evolutionary brain" comes from evolutionary algorithms. As I was exploring the world of AI, I eventually came upon some videos by a graduate student named Julian Togelius which just shocked me. Here was a car learning to drive on its own. This convinced me of the power of evolutionary learning algorithms and I became intrigued by the idea that this could be the basis of learning in human intelligence.

I had been thinking about creating a blog on this topic for sometime, but I was concerned that it might turn into a time sink and so never actually went through with it. The thing that finally convinced me to write my thoughts online was an article by Kevin Kelley titled "Evidence of a Global SuperOrganism". It was so thrilling to find someone expressing ideas similar to my own about AI. It seemed like his ideas would parallel mine and then slightly diverge and then come back in unison. I soon learned the KK was in the process of writing a book about what he calls the Technium, and that his blog was a way of exploring the ideas that would eventually become his book.

So I decided to finally create this blog to provide a public space to write my responses to Kevin Kelley's Technium. The highest honor for me would be if KK himself would come visit my blog and incorporate some of my ideas into his book.

This will also be a place for me to discuss AI in general, and to discuss my own Evolutionary Brain Project in particular. I expect that the topic of the Singularity will come up fairly frequently. I have amassed a huge number of links on AI in my informal studies of the subject and I might use this blog to highlight some of those from time to time as well.

And so this blog is born and begins its own process of evolution.