![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
unbibiumI've always said that computers are stupid.
Meaning, they don't know anything of consequence. They have the ability to do lots of math, and move bits around, and even process certain kinds of input and output in ways that humans can sometimes understand. But there will always be a communication gap between humans and computers, and an abundance of confusing bugs, as long as computers don't possess common sense.
Thankfully, I'm not the only one with this idea, as someone has been building a computer with common sense named Cyc. I was surprised to learn that a computer asked whether it was human in 1986. Did that make it in the newspaper? It'd be a shame if it didn't, what with glorified chess-playing algorithms getting all the press recently.
Apparently it's now accessible on the Net, and you can teach it things. But that worries me a bit, since I just know a bunch of people are going to try and teach it things that are either wrong or extremely biased. Hopefully Cyc knows how to distinguish fact from opinion. I also wonder if people will try to pull a Captain Kirk on it and attempt to argue it into self-destructing, like he did in at least three episodes of Star Track. That being said, I wish I had Linux installed so I could experiment.
But I can't help but wonder how culturally biased he is to begin with. Would a British Cyc have disagreed with the American Cyc on some key assumptions? What would a Japanese or Arab Cyc take for granted that the Russian or African Cyc would dismiss as just wrong?
For the above reasons, I'd feel better about Cyc's usefulness if he is given the ability to consider his sources, and perhaps even cite them. This would be useful, for exmple, when one asks him who killed Jesus. "Historians say X, but Baptists generally believe Y..."
It's encouraging that they've taught it curiosity and tact; I wonder how long it will take to understand other behaviors and emotions, if it doesn't already. Maybe in a few years, he'll be able to understand jokes, since most of them are based on violations of assumptions.
Enjoy it while you can, as humanity is obviously doomed once he reaches self-awareness.
Meaning, they don't know anything of consequence. They have the ability to do lots of math, and move bits around, and even process certain kinds of input and output in ways that humans can sometimes understand. But there will always be a communication gap between humans and computers, and an abundance of confusing bugs, as long as computers don't possess common sense.
Thankfully, I'm not the only one with this idea, as someone has been building a computer with common sense named Cyc. I was surprised to learn that a computer asked whether it was human in 1986. Did that make it in the newspaper? It'd be a shame if it didn't, what with glorified chess-playing algorithms getting all the press recently.
Apparently it's now accessible on the Net, and you can teach it things. But that worries me a bit, since I just know a bunch of people are going to try and teach it things that are either wrong or extremely biased. Hopefully Cyc knows how to distinguish fact from opinion. I also wonder if people will try to pull a Captain Kirk on it and attempt to argue it into self-destructing, like he did in at least three episodes of Star Track. That being said, I wish I had Linux installed so I could experiment.
But I can't help but wonder how culturally biased he is to begin with. Would a British Cyc have disagreed with the American Cyc on some key assumptions? What would a Japanese or Arab Cyc take for granted that the Russian or African Cyc would dismiss as just wrong?
For the above reasons, I'd feel better about Cyc's usefulness if he is given the ability to consider his sources, and perhaps even cite them. This would be useful, for exmple, when one asks him who killed Jesus. "Historians say X, but Baptists generally believe Y..."
It's encouraging that they've taught it curiosity and tact; I wonder how long it will take to understand other behaviors and emotions, if it doesn't already. Maybe in a few years, he'll be able to understand jokes, since most of them are based on violations of assumptions.
Enjoy it while you can, as humanity is obviously doomed once he reaches self-awareness.
![[identity profile]](https://www.dreamwidth.org/img/silk/identity/openid.png){kind=small}
no subject
Date: 2002-06-07 12:46 pm (UTC)(No, it's not ironic or sarcastic. I want computers to be human. That's the way I'm wired... ahem.)
Have you read
Date: 2002-06-07 01:06 pm (UTC)Re:
Date: 2002-06-07 03:11 pm (UTC)Though MegaHAL is the most fun to try to talk with. We installed an IRC bot with MegaHAL and put him in the #Kibology channel, and nobody ever had a human conversation there again. We just kept feeding gibberish to MegaHAL to see what he'd say back.
no subject
Date: 2002-06-07 03:48 pm (UTC)