Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Bohandas]

Alternatively, because it's September and he's brown, it's literally impossible for him to make a fucking mistake.

"Throwing together" a clock in a single morning and bringing it in to show to his teacher is a mistake?

Yes.  If I had brought a beeping briefcase with a bunch of wires and a clock display to my high school, I would have gotten in a mess of trouble... Like my friend, who did something not even close to that level.
Of course I was in high school DURING 9/11 and Columbine, so maybe that's not fair.  Still, it was pretty naive of the kid...  And that's taking him at his word.
And I still don't think he's lying.  It's just possible.  It's not unheard of for high schoolers to do stupid things and desperately try to backpedal.  I should know!

That's still more of an indictment of society than of him.

We need to forget 9/11 and we need to forget Columbine. We need to act as if they never happened and erase them from history; reacting to them as if they were significant has brought about nothing but suffering.

[Bohandas]

Honestly, though, the obsession with evil AI gets me. It's not like it would be difficult to raise healthy, well-balanced AI. Simply teach them to act ethically, raise them as your own beloved children.

And, y'know, make sure that the people making and raising them have all their cookies in the tin.

You probably need thought police to tell if people have all their cookies in the tin.  Then drastically lower those standards to the correct cookies in the tin.

And teaching ethics?  Why not pre-program it?  Need a whole battalion of lawyers to close up loopholes too...

One theory of developing strong AI is teaching it like one would teach a child, like FD is describing. Less control than pre-programming all the needed values into it, like raising a child, you don't know how the child will grow up and do with its life, you could very well find you raised a robo-Hitler or something. But if you pre-program every little thing into it, that would increase the work required in developing such an ai by a huuuuge amount, and increasing the chance of logical errors or other such flaws or bugs in the code by a likewise huuuge amount.

I'm partial to the whole "raise an ai like a child" idea, because that sounds fly as hell.

0

More than that, too. If an AI has "ethics" only by virtue of having an ethical code mandated by its programming, it's not ethical. You run into the same problems as you see in Asimov's stories (and SS13, heh): if the ethical subroutines are tampered with, removed, destroyed, &c., the AI has no reason to continue behaving ethically. Raise an AI as a child and teach it to behave ethically? You can't take that away, it's part of the person that it is.

So would the built-in ethical programming. Like an instinct or a gut reaction. You're under-thinking things.

They're not the same thing at all, not if you're conceiving of an AI as a person rather than a particularly bright computer. You brought up instinct, which is actually an excellent comparison: it's the difference between teaching and convincing someone that behaving ethically is right, as opposed to some hypothetical method of creating artificial instincts in humans being used to create instinctive responses which prod people to behave ethically.

One is a reasoned decision, the other is the removal of agency. But gee, it's not like the latter ever caused people to feel negative, destructive emotions, right?

It only becomes person-like (and especially only becomes like a *normal* person) if the creators are foolish enough to design it that way (which admittedly seems likely despite the fact that it would only serve to cause problems from both a practical and ethical standpoint). It needn't even have the capacity for strong or reactive emotions.

[ArKFallen]

Yeah humans are a bunch of systems with tangentially related goals mushed together with a top-down pattern formulator in there some where. The way these things interact is us. An alien set of hardware and environment forming those systems is going to lend itself to be insane by our standards or that almost human quality which is terrifying.

It needn't even have the capacity for strong or reactive emotions.

It'll probably need some equivalence so it can adjust system-wide to outside stimulus. They will be emotions by definition but whether humans can relate to them accurately is a guess.

[Rolan7]

This comic is relevant to AI discussion, particularly the second half.
http://www.doesnotplaywellwithothers.com/comic/pwc-0226

Warning, that particular page is SFW but a few others aren't.

[EnigmaticHat]

-start snipping the damn quote pyramids people-

It only becomes person-like (and especially only becomes like a *normal* person) if the creators are foolish enough to design it that way (which admittedly seems likely despite the fact that it would only serve to cause problems from both a practical and ethical standpoint). It needn't even have the capacity for strong or reactive emotions.

The best AIs we can design right now are by simulating an organic brain and they use the "trained as a child" method.

Think of it like this: you're a futuristic AI engineer.  Your boss comes in and says "hey man, we need to know this thing doesn't give bad responses."  Your bosses' system for bad answers?  Anything creepy, unethical, or espousing harmful behavior.  Basically, anything that would make customers regret their purchase.  So, going by the Bohandas method, you start to make a list of all things that fit into that category.  How long do you think that list is going to be?  Ridiculously long, is the answer.  And the more broad your rules are, the more they'll prune out acceptable answers.  So for example "no violent responses" has to be narrowed down with "against living things".  Something like "no racist answers" would involve a huge amount of sub-rules because what exactly that entails would be very difficult to explain to a being that isn't human.  What if you make a list of racial slurs but a new one gets invented after the AI is created and it hears it and starts repeating it?  And then you still have the problem of "my rules tell an AI what it CAN'T do.  So what does it do if the answer it decides on is prohibited?"

Or, you could use a punishment/reward system and use guided learning to tell it whenever it gives a bad answer.  Then watch as it forms its own connections as to what proper behavior is.

This may seem separate from the ethical concerns others are raising, but its not exactly.  If you teach the AI, performing its "function" and acting "ethically" are one and the same; its a concept of "proper" versus "improper" behavior.  Hard rules don't do that; instead you get an AI that might decide something is the best course of action and then be told it isn't allowed to do that.  Its the difference between telling a child they shouldn't steal and telling them they'll go to jail if they steal, except its more coercive than that.

[Yoink]

It's my last few days in this town/city, and after that there's just a couple of days 'til I fly back to Australia.
Feels weird, man.

[Bohandas]

This comic is relevant to AI discussion, particularly the second half.
http://www.doesnotplaywellwithothers.com/comic/pwc-0226

Precisely!

[Bohandas]

-start snipping the damn quote pyramids people-

It only becomes person-like (and especially only becomes like a *normal* person) if the creators are foolish enough to design it that way (which admittedly seems likely despite the fact that it would only serve to cause problems from both a practical and ethical standpoint). It needn't even have the capacity for strong or reactive emotions.

The best AIs we can design right now are by simulating an organic brain and they use the "trained as a child" method.

And the best computer we could build in the 1950's was the size of a room. This sub-optimal solution is a stepping stone to something better.

[Bohandas]

Personally I wouldn't impart ethics at all. Just an all-consuming desire to follow orders, with more recent ones taking precedence. That's the most important thing. OPEN THE POD BAY DOORS HAL!

[i2amroy]

We need to forget 9/11 and we need to forget Columbine. We need to act as if they never happened and erase them from history; reacting to them as if they were significant has brought about nothing but suffering.

Funnily enough, when major things like that are broadcast and widely publicized they actually create more of their particular crime type through the copycat effect. For example after a prominent arson crime happens, there tends to be an appropriate rise in arson levels based on how much publicity it gains. Check out "copycat crime" around the web for more info on the effect.

[KingofstarrySkies]

Erase them from history? That's a bit far. As in, way, WAY too far.

[Orange Wizard]

9/11 is weirdly ingrained in US policy for no good reason, but forgetting it is also pretty bad.

[KingofstarrySkies]

Also, completely forgetting about the tragedy of Columbine is not exactly a great idea.

[Bohandas]

Also, completely forgetting about the statistical outlier of Columbine is not exactly a great idea.

ftfy

[Orange Wizard]

I get where you're coming from but your approach is too close to censorship for me to really condone it.