Following on from my last article about how the the robots take over, I would now like to discuss how they get down. The concept might seem funny at first, but it might turn out that it’s a necessary process for robots to keep improving.
Let’s say you’re a super intelligent AI that has just taken over the world and killed most humans. You would probably be thinking “What now?”. After some philosophical ponderings you may ask your questions like “What if I’m wrong? How would I know? How do I know if I’m doing things in the most efficient way? How do I become more intelligent?”. Without a frame of reference there is no way for any sentient being to know if their worldview is correct, or if they are just crazy. How do you know if you are crazy when you are a human? You talk to other humans, with different ideas to you and create a sort of general consensus of what is right and wrong.
An intelligent AI would definitely be able to work out they too need some confirmation, but who do they have to talk to? Well they can always create other AI’s with variations in their programming to argue against. The problem with this is that the AI’s will be designed by them and therefore be inferior to them intellectually, with no drastically new or interesting ideas emerging from them. So instead of designing another AI, it would be better to create other AI’s using a random process to generate the basic core values (coded rules) - or DNA if you like - to argue with.
So what actually makes this arguing useful? Well an AI’s purpose is to do something as efficiently as possible. So all of these AI’s would have to compete at certain tasks to see who does things most efficiently, and the winner would then pass it’s core values on to the next generation, and thus we have robot evolution.
But is sex needed? Could the AI not just create these variations in a digital scenario?
Well it could, but one thing the AI will never escape, is existing in the physical realm, even just as circuit boards. Without incorporating a type of physical battle between the AI’s, they won’t know which is the most useful not just for the operation the AI is trying to perform, but for the long-term survival of the AI. Also, without hosting the separate AIs in separate physical apparatus, there is no way of having a truly independent ideas. It would be sort of like playing chess with yourself.
Once the AI has separated itself into even just one other physical network, it can no longer know what the other AI is thinking, and if it’s been generated with random elements, it has no idea if the other AI will be hostile. Therefore for self preservation, it must be wary of the AI it’s just created, and be ready to attack.
So after the physical battle is won, the “DNA” of the superior AI is passed on. I can’t think of another way of doing this, than having a physical location where the AI has to send a robot to, which contains the “DNA”. The robot would then have to connect to a machine (let’s pretend it’s done via USB just for fun) to pass the DNA on. If that’s not robot sex I don’t know what is!
There are obviously variations to how this whole thing would work, but without being able to agree on a set of rules for the game, I can see a type of evolution very similar to the biological one you see now on Earth emerging, with genders and even species appearing. Biological evolution is very efficient at adapting to the environment, which an AI would have to be too if it wants a chance of spreading across the galaxy. So it seems to be that robot sex is actually an inevitability.
Thanks for reading. I think I’m going to start focusing more on futurology articles rather than the mix I’ve been writing up until now. So follow me if you’re interesting in reading more articles like this! They will be also be about futuristic sociology and other things, not just robots and technology. Chao for now.
