Deux Ex Machina, God in the Machine (C) (O)

Jimmy's recent post on Babel got me thinking about a modern day version of the same principle. The technological "Singularity". I'm fascinated by it, a junky for new information on it. I first heard of it in a science fiction short story I read in high school, it's been a brain hobby ever since. (I have lots of brain hobbies, things I like to think about and follow along with, it doesn't require any physical labor.)

The Singularity, sometimes called the Technological Singularity, is a theoretical point in time where artificial intelligence (AI) passes human intelligence. At that point, the intelligence driving the creation of new AI is no longer human, but machine intelligence. In other words, we create machines that gain "intelligence" and at a certain point in time those machines have an intelligence superior to ours, at which point they "take over" the job of driving the further development of the craft. What happens after that is pretty hard to say, but it happens very fast.

I'm sure at this point alot of you will put this in the category of science fiction, so let me explain something.

Artificial Intelligence is a term used to describe two things. First, the intelligence of machines. Intelligence of machines being defined as ability to learn, reason, judge, communicate, etc. All the same things we associate with human intelligence. Second, the branch of computer science which works to create it. At this point in time, 2009, there is a lot of money in AI and most of the technology being developed in the world today would be impossible without it. Almost everything humans do with computers, from medical diagnosis, stock trading, video games, everything down to the machine that dispenses coupons of the shelf at the grocery store uses AI.

The field of AI is based on the idea that human intelligence can be described so precisely that it can be explained to and then simulated by a machine. Go to a grocery store and pull a coupon out of one of those dispensers on the shelf. A new coupon comes out. Pull the new one out, and another one comes out. Pull that one out really fast, and while the new one is on it's way out, pull it out too. The machine stops, waits for a minute, and then puts out a new one. This discourages kids from grabbing a million coupons for no reason. That's one example of AI I learned about when I was seventeen. Program every single possible chess move into an application, assign value codes from low to high for the strategic implications of each move, and then let the application play against a human. Those applications can now beat Kasparov. That's a more complicated example of AI.

Computers using AI applications can now paint painting and write novels. The paintings and the novels are not very good yet, but they're getting better. I was at a Mensa group one time where a guy would read two short stories, one written by an application and one written by anyone at the meeting who wanted to write one. I wrote one of my own. The guy would read the story you wrote, and the one the machine wrote. 500 word or less stories that had to have a definable protagonist, antagonist, beginning, middle, and final act. Most of the time he was right, a couple of times he got it wrong. That was four years ago.

Point being, computers can already "reason" and "create". We've taught them that.

We can teach them anything. Even common sense.

A human brain, even one in a very "stupid" human being, contains billions and billions and billions of pieces of information about as big as a line of code. Just the portion of our minds we would call "common sense" would amount to trillions of lines of code. Each line has to be entered, one by one, and each line has to receive a value judgement based on some kind of agreed upon list of rules for what could be considered "good" judgement.

A human being sees a green light but also sees a car that isn't stopping, so they don't step onto the street. In human terms that is very simple, in AI terms that's incredibly complicated. I don't want to beat this point to death because it isn't really the big picture, but for a human brain to identify something as a car is a million pieces of information, to identify speed and distance and force is a million pieces of information, and to perceive a risk that is either too high or too low and run that risk against a list of lessons learned in life, and then decide to NOT WALK, requires the use of tens of millions of lines of code.

The thing is, that's all it is, is: code. The idea behind the singularity is that at a certain point in time, the AI itself starts to develop faster ways of getting that knowledge out of our heads and into digital bits of information the machine can understand.

When we reach that point, we no longer have any real idea what happens next.

Think of it this way, you teach a kid how to ride a bike. First the kid has to get old enough to push the peddles and steer the wheel, then you have to explain it to him, then he has to try it over and over until something clicks and then you don't have to push him anymore, you have to chase him.

It's the same with AI except that at some point it takes off on it's own and when it does, we have to assume it changes course altogether, because of the rate at which it changes. If it takes a team of humans a year to enter a billion lines of code into a computer application, once the application becomes smart enough to type it's own source code, it might enter a billion lines of code in an hour. And the more code it feeds itself, the faster it can enter new code. And the more new code it enters, the faster it can feed yet more code. So what took a team of human beings twenty years to do suddenly becomes something the application can do a thousand times per second.

That's the Singularity, and once that moment in time comes, the machines start to build new machines on their own. For what purpose, to what end, we know not.

When you have a world with a superhuman AI, along with nanotechnology, you have a world where a human being is really quite superfluous. Once they've learned everything we know, "they" don't need us anymore.

Don't try to imagine what they would think or what they would want. You CAN'T imagine it. You're not smart enough. Whatever you think is just one through eight on a list of a billion things the application thinks in a nanosecond. It knows everything you know, and everything you don't, and it can think exactly the way you think, or better than you could possibly think, if it wants to.

Reasoning, learning, perception, philosophy, etc., are all basic subroutines. Socrates' life's work, every thought he ever thought and put into words, is the turn of a switch for the application.

Take a million Einsteins and put them in a room together, give them five million years to work on a problem like cancer, greenhouse gas, renewable fuel, SIDS, you name it. The application can simulate that a hundred times a second.

Now here's where it really gets interesting...

A human brain is a very sophisticated computer, but its just a computer. Every thought you think can be expressed in zeros and ones. A machine can "read" your brain, can download the information in it, so to speak. There are already applications capable of this. But when you throw a superhuman AI into the mix, you're no longer talking about taking information out of your brain and translating it to code, now you're talking about taking lines of code and translating them into human intelligence, and conceivably putting it BACK in a human brain.

There's no need to worry about how much information can fit in a human brain. Think of every single piece of information in the world as being written down on a single sheet of paper that stretches to the sun and back five hundred times. You can break that list into sections, as many as you need, and store different sections in different locations, and access whichever section you want, as needed. So you don't need to fit that whole list into your brain, you just need to fit a few access coordinates and a basic dictionary.

So with a superhuman AI, you yourself have a brain that can "know" anything you wish to know about anything.

And now the really juicy questions start.

Is your "mind" an actual physical thing, or is it actually a RESULT of all the information in it?

Is there even such a thing as intelligence, or is what we now call intelligence just a crude method for collecting and arranging information?

Is a human being a body, or is it a state of mind, and if a state of mind is something that results from a particular quantity of data, who is to say at what point the machine becomes as human, or MORE human, than a human?

Is Love code?

Once you erase the need to pass your knowledge onto a biological offspring to ensure the survival of the species, is there a need for love?

Is art code? Is the appreciation of art code, or is the appreciation of art a biological response, and therefore an unnecessary response. A superhuman AI could create a work of art at least as great, if not far greater, than anything DaVinci or Mozart ever did. It wouldn't need to, but might it want to?

Is the desire to "create" a uniquely human desire that hearkens back to evolution, or is it the next logical step once you've gotten a basic grasp on systems?

The list goes on and on, it's all too much. And after the Singularity, we find all of this out as quickly as our brains can comprehend the information.

People who look forward to and follow our progress toward the Singularity are sometimes called Singularists, I proudly call myself one, and the leading thinkers in the field are projecting a date of between 2010 on the early end, and 2040 on the latest end. My money is on 2030 ish.

Take a superhuman AI, couple it with nanotechnology and bioengineering, and you've got a human being who can live forever. The body you live in right now can live forever if a superhuman AI can arrest the aging process, and if it can't it doesn't matter, because you can clone a new body and move your brain into it. You could be twenty five forever. Or thirty five if you prefer. Of course it's all in vanity, because you don't even need a body. You can exist in lines of code that can go anywhere, and are everywhere at some point. (You can store data in any computer, and you can make a computer out of blades of grass, you can make a computer out of black holes in the galaxy.)

No more disease, no more suffering, no need for human labor of any kind. (Sound familiar?)

If you've read this far, and if the Singularity is a new idea for you, you're probably thinking about the religious implications. What does the Bible have to say about superhuman AI? How does an immortal mind that doesn't even need a body fit into a Biblical notion of life and death? Will God allow humans to develop AI to this extent? I think alot of people would say no.

But I think that's a mistake.

A big mistake.

Five hundred years ago the Catholic church thought a belief in a round earth was blasphemy. A hundred and fifty years ago scientists thought Darwin had finally "disproved" the Bible. There are now more Christians alive on earth today than there were human beings alive on earth at the time, and all those scientists are long in their graves, and many Christians now use the theory of evolution as a tool to explain who God is.

People living on a farm ten thousand years ago would look at the world we now live in with the same wonder and awe and confusion that we would look at a society twenty years past the Singularity. Whatever notion those ancient farmers may have had about our 2009 Earth is no more and no less ignorant than our notion of a post Singularity society.

We're cavemen trying to imagine what the dress code inside NASA might be. We shouldn't do that. because we can't help but be wrong.

My hope is that I live to see it happen.

Stem cell research is part of the process that ultimately ties our brains and our bodies to the machine. And we're still talking about embryos that won't ever be used for anything else anyway.

So if you disagree with me and think we shouldn't do it, I am placing the ball back on your side of the court.