der_kluge
Adventurer
Not all were bad. You forgot Robocop!
You know, despite advancements in computer science, we're still decades away from computers that can think for themselves, if not at least another century or two. So, we don't have anything to worry about for quite some time.
Besides, most artificial intelligence is so limited in scope that the idea of an "I Robot" is a bit far fetched. We have robots that paint cars, or ship packages, or whatever, but that's their entire world, is just that specific task. I just can't think of any good reason why it would be necessary to try to program a computer to know *everything*.
There was a team trying to do that once. Don't know whatever happened to it. They were keying in "basic facts" into a program. Things like, "umbrellas aren't needed indoors" or "buses carry kids to school". Just basic stuff. Thing about all the *things* that you know about life that you take for granted, and then attempt to teach all that to a computer. It's why a lot of research has focused on "learning machines". Just program a computer enough to be able to observe and learn on its own, and it can build it's own database of knowledge.
I had an idea in college to teach a computer to read the dictionary. Start with say, "aardvark" and recurse through words until it resolved a stack of definitions. In other words, give it some basic knowledge of certain concepts to get it started. Then, have it read the aardvark entry. So, it would be something like "an animal...", so it might not know what the word "animal" is so it'd flip over to that entry to get that definition. The definition for animal might be "a living..." so it might not know what "living" means so it would flip over to that definition to understand that word. And so forth. Eventually, it would come back to the word aardvark (several thousand recurses later) and get to the next word in the aardvark dictionary.
You know, despite advancements in computer science, we're still decades away from computers that can think for themselves, if not at least another century or two. So, we don't have anything to worry about for quite some time.
Besides, most artificial intelligence is so limited in scope that the idea of an "I Robot" is a bit far fetched. We have robots that paint cars, or ship packages, or whatever, but that's their entire world, is just that specific task. I just can't think of any good reason why it would be necessary to try to program a computer to know *everything*.
There was a team trying to do that once. Don't know whatever happened to it. They were keying in "basic facts" into a program. Things like, "umbrellas aren't needed indoors" or "buses carry kids to school". Just basic stuff. Thing about all the *things* that you know about life that you take for granted, and then attempt to teach all that to a computer. It's why a lot of research has focused on "learning machines". Just program a computer enough to be able to observe and learn on its own, and it can build it's own database of knowledge.
I had an idea in college to teach a computer to read the dictionary. Start with say, "aardvark" and recurse through words until it resolved a stack of definitions. In other words, give it some basic knowledge of certain concepts to get it started. Then, have it read the aardvark entry. So, it would be something like "an animal...", so it might not know what the word "animal" is so it'd flip over to that entry to get that definition. The definition for animal might be "a living..." so it might not know what "living" means so it would flip over to that definition to understand that word. And so forth. Eventually, it would come back to the word aardvark (several thousand recurses later) and get to the next word in the aardvark dictionary.
Last edited:
