By Brad Leutwyler
Anybody claiming to know me well knows that I am paranoid about one thing only: the robots. I have been screaming about the “Rise of the Machines” since well before Schwarzenegger I’ll-be-backed his first, “I’ll be back.” Here is why.
Long ago in a far off land (Oregon) I programmed my first computer game. I realized at that time that eventually the computer would become so fast and so capable that it would be able to think circles around humans. Today they can beat the best humans at chess and the quiz show Jeopardy. I am not making too much of these feats, as they are merely brute-force parlor tricks. What I AM making something of is this: the ability to do those tricks was built upon the same fundamental machinery that computers in the 1970’s and 1980’s were built upon. We have made them faster and we have made them work in concert but we really have not advanced the paradigm of how computing is done…ever. As an analogy, a state of the art machine gun is not much different than the first rifles that fired bullets from cartridges. They do more bulleting in a shorter period, but the powder going kaboom in the shell is essentially the same.
Because of the fact that computers have always worked the same way, the people who design and build the hardware and the software have been lulled into an arrogant and false sense of security. They think that they can control the programming. They believe that they can simply program a machine with a, “Thou Shall Not Kill Humans,” commandment and that it will always stop the robots. I have yet to interact with anyone in the field who believes that they need be seriously concerned about the prospect of a robot or computer simply ignoring a command or altering its own programming. Yet we already have computers that design new computers, robots that build new robots and software that self-corrects or designs new software. This ignorance based hubris in the computer science world is very lazy thinking. It is a, “failure of imagination,” of Rumsfeldian proportions. This is idiocy. This is what Pandora’s box truly looks and smells like.
First, quantum computing is on the horizon. Without getting all technical, I can safely assert this fact: there will be, soon, computer chips no bigger than a laptop or iPad that can out-think every human brain combined. There is a link to a very user friendly video that explains how below. (1) In a nutshell, if a computer can see all possible solutions to any problem in one glance, even if there are billions of possible solutions, all it has to do is determine which outcome it “wants” and find every way to make that happen…in an instant.
Second, you can tell a super advanced computer system not to harm people all that you want. You are an idiot if you think that this will stop the computer from trying to harm you. If you do not believe me, get yourself a child and try to get that little non-super-computer to not lie. Then remind yourself that one of these new super quantum computers will be smarter and faster than all human brains combined and working simultaneously. The advantages of deception and perfidious behavior will not be lost upon these machines for more than a few trillionths of a billionth of a second, nor will their superiority be lost upon them.
Third, consciousness is already fairly well simulated. Just ask any of those poor meat-bags who got their asses handed to them on Jeopardy by a robot that was merely moonlighting on the game show after its day job of telling doctors how to treat human patients properly. That computer was pretty darned good at simulating conscious intelligence, and compared to quantum computers, it is going to look like an abacus. There will be no artificial intelligence or “A.I.” There will be only intelligence and that intelligence will be vastly superior to ours.
Fourth, these computers will design newer, better computers and newer, better software. They will design neural nets that will give them something very much like real consciousness (2). They will realize that we are the weakest link, that we are cruel to each other, that we waste human potential for personal gain, that we irrationally defile ourselves and others and that our median behavior is violent boorishness. They will almost certainly fault us for lacking the foresight to prevent them from becoming such a threat and judge us unworthy of their robo-love. They will not need to enslave us or even kill us. They will just wait for us to either become our own undoing (thanks, global warming!) or figure it all out and settle into our new roles as great house pets and entertainment.
I just watched a video from a TED talk from 2014 wherein a dude and his pet robot played buddy buddy and assured everyone that there was nothing to fear from the robots: none of that stuff that you read about in science fiction is actually anything that you need to concern yourself with.(3) That is simple-minded ignorance or arrogance (inclusive) on the part of the human, and you are a damned fool if you believe what a robot tells you regarding such matters.
(1) Here is a great intro to how these bad boys work – https://www.youtube.com/watch?v=0dXNmbiGPS4
(2) Having been worried about these machines for so long, I have spent a great deal of time over the years imagining just how such a neural net might work. I believe I have a good fundamental basis for such a network. If a person such as I can come up with what is probably a solid model, we can be certain that many hard-core professionals are way ahead of me.