Author Options:

Database of facts project : Teaching Good and Bad to a machine ... Answered

Let's imagine that you successfully created an intelligent machine, capable of thinking.

The purpose of this machine would be to replace all of our governments and judges, so the world become a better and fairer place for every of its inhabitants.

It would be an IA managed globalization.

As you've just created this machine, her brain is empty. Like a tiny baby, she has to learn a lot ...

Question : what facts would you teach her ? (given that your goal is to make her a fair judge and good world leader)
What would you teach her to make her able to make the difference between what's good and what's bad ?


I might first teach it the dictionary...then have it download wikipedia.


Why not teach it the unimportance of gender in a true meritocracy?

I think it's a french thing...aren't all nouns either masculine or feminine in french?

Yes they are. AI and database are female in French. Computer and algorithm are male. However, I said "her" mainly because I wanted to keep the possibility to seduce her ... just in case something goes wrong with her algorithm ...........

My, what lovely LEDs you have. Are they natural, or are they doped?

And there would be some quite complicated "logistical" issues. Connectivity and the like...

.. Computers are either malevolent males or feminine females, dontcha know? You and your silly egalitarian ways.


10 years ago

The computer scientist in me wants to say
"0 exists. Every integer has a successor. Have fun."

..but I don't think that's what you're after, is it. If you mean facts that would make him/her/it able to make judgements about the real world, I have absolutely no idea where to start.

Actually, the rules could be the same "rules" you'd teach to your own children. =o)

What would be the first rule you'd teach to an intelligent machine (or to your creature) ?

Well I think a lot of the first things I'd teach my children would be along the lines of "don't put that in your mouth" :P If we're talking rules, then the categorical imperative, "be nice to people", sometimes even a bad thing is the right thing to do, "no nation ever benefited from a protracted war", etc...

. Wow! This ought to get interesting.

This would be very difficult. What is good in one context is not always so in another. If it had to reason from a set of strict rules, like most computers must, it would fail to be just in all cases, and maybe in most cases. If you want to explore some of the things that would happen if given strict laws to live by, read some classic short story anthologies from Isaac Asimov like: "I, Robot" "The Rest of the Robots" "The Complete Robot" & "Robot Dreams". They are literally a compendium study on the effect of laws on machines interfacing with humans. "Liar !" it both one of my favorites and also one of the saddest of the short stories. Bicentennial Man was one of the most interesting to me.

Bicentennial Man was one of the most interesting to me.

I love that movie.

I'm actually surprised that someone else has seen it....

Hmm, yes the movie was very well done. Even followed the actual story pretty well. But I read the story over 30 years ago, in one of the anthologies I mentioned on robots ( "I, Robot" I believe and; BTW the movie "I, Robot" did not follow the book named the same in ANY way at all)

. I agree that it is next to impossible to do, but I think it will be interesting to see what ppl come up with.
. Asimov rocks! The Robot Series is definitely required reading for this discussion. The Laws of Robotics work very well, but, as you point out, no algorithm is perfect for this.

> but I think it will be interesting to see what ppl come up with.

Hehe, this is precisely what interest me the most actually.