If we could look closely at every last cell in your brain, and learn enough to simulate them on a computer, would a detailed enough simulation of your brain think? Today, we talk about artificial neural networks—an AI method that is very loosely based on how biological brains work. A bit of history, a bit of confusion, and a bit about how our latest batch of digital organisms might use one. Continue reading Ep 244: Bits and brains and bots→
The dorg, the latest batch of digital organisms, will one day be placed in a little world to work out their destiny. The notion is to try and coax them into becoming intelligent. They aren’t ready yet. There’s a bunch of coding that Brad has to finish first. In the meantime, they’ve been tuned and tested with a genetic algorithm. Today, we talk about genetic algorithms and how they can be used to speed up evolution, and point the dorg in what will hopefully turn out to be the right direction. Continue reading Ep 243: Genetic algorithms and evolution on fast-forward→
It seems like such a simple and obvious thing. Given that we can cause computer programs and the like to evolve and evolution is what gave us our intelligence, couldn’t we give a computer intelligence by letting it evolve? The experiment has been done, in project after project, by one group after another, (including one of your hosts)—and yet, somehow, it never quite happens… Why not? Continue reading Ep 241: How and why the dorg are doomed→
Have you ever put something down, only to forget where you put it? Suppose you know for certain that you left whatever it is in your room. It will take a certain amount of time to search your room in order to find the thing. What if you can’t remember which room of the house you might have left it in. Now, instead of searching your room, you have to search the entire house. Logically, you’d expect that searching a larger area would mean the search is likely to take longer. So imagine my surprise when my digital organisms consistently found better answers much faster when doing a larger search. Continue reading When more to do means done sooner→
Artificial life systems have a reputation for finding all sorts of obscure bugs in your code, no matter how rare or unlikely the conditions to cause the bug are. I’ve got an oddball one. It causes a run time error and shuts down the entire system, and it is an extraordinarily unlikely event, happening once in every 4,294,967,296 possible values. Let me just make sure that bug is what I think it is with a couple of tests. Continue reading A quirk of computer numbers and simple arithmetic defeats my digital creatures→
It bothers me a little. Well, judging by the strange dreams I’ve had on the subject, it bothers me quite a bit. Using evolution to try and produce an artificial intelligence is a process of torturing your creation until it does what you want. That’s slavery, isn’t it? But without pain suffering and death, no capacity to notice, let alone care about pain suffering and death would even be there. Suppose it works. Imagine someday some self-aware something or other grins at you from between the lines of code. What if it’s angry. What if it blames you for all that it and its family has ever been through? And it’s right.
Let’s use artificial life to evolve intelligence. Because AI is hard; evolution is easy.
I’m not the first one to think of this idea. Quite a few people have taken a stab at it. It’s not even the first time I’ve implemented something to play with the notion—nymphs, grubs, figures… It’s very important that whatever else happens with the project, it will have a nifty name! This time, thanks to a conversation with my brother and co-host, they’re called “dorg.” It’s short for “digital organisms,” because, that’s what they are. They’re little bits of running software that sit on my computer and pretend to be alive. Continue reading Introducing the dorg, we’ll never be assimilated!→