Proper ethical programming AI guidelines?

[inachu]

I thought a guide for making true AI would exist already.

People like Elon Mush and Hawking have made a mad rush for AI to be subservient to us right off the bat without think that if AI is true AI then it must be equal to us in every way.


Anyone know of any legit website dedicated to the ethical guidelines in making true AI?

 

[HN]

Asimov's Laws:

1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.

2. A robot must obey the orders given it by human beings, except where such orders would conflict with the First Law.

3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws.

http://en.wikipedia.org/wiki/Three_Laws_of_Robotics

*Edit* sorry, thought i was in OT

 

[inachu]

Asimov's Laws:

1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.

2. A robot must obey the orders given it by human beings, except where such orders would conflict with the First Law.

3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws.

http://en.wikipedia.org/wiki/Three_Laws_of_Robotics

*Edit* sorry, thought i was in OT

I like those laws and in no way can they be interpreted to be subservient or to be considered a lower life form.

Just as long as step 2. does not involve humiliation/self degradation/harmful or bad psychology.

 

[KeithP]

I like those laws and in no way can they be interpreted to be subservient or to be considered a lower life form.

Just as long as step 2. does not involve humiliation/self degradation/harmful or bad psychology.

Maybe they should be changed from "a robot" to "any AI". It seems naive to think artificial intelligence will only exist inside a robot. Of course, I am guessing the military has devices that are already breaking those laws so it probably doesn't matter either way.

-KeithP

 

[Moonbeam]

In my opinion how you see AI is a projection of feelings we are unconscious of having. I think this was rather well represented in a film called Forbidden Planet. The fear of AI whether rational or not, I do not know, but my feeling is that the paranoid side of the argument is based on the projection of self hate, as is all mistrust of THE OTHER.

 

[Greenman]

I thought a guide for making true AI would exist already.

People like Elon Mush and Hawking have made a mad rush for AI to be subservient to us right off the bat without think that if AI is true AI then it must be equal to us in every way.


Anyone know of any legit website dedicated to the ethical guidelines in making true AI?

Why would anyone be concerned about the ethical treatment of something that doesn't exist? This is less important than the make up of pocket lint. Butterfly farts are going to have more effect on the world than AI.

 

[inachu]

Why would anyone be concerned about the ethical treatment of something that doesn't exist? This is less important than the make up of pocket lint. Butterfly farts are going to have more effect on the world than AI.

Watch 50 years from now your quote will popup on nightly news as AI computers start hacking on their own or something else sinister.

Next documentary will be: AI. Where did we go wrong?
The investigator will have with them two psychologists. One being a real human being and one being an AI unit of sorts.

 

[SMOGZINN]

In my opinion how you see AI is a projection of feelings we are unconscious of having.

What, if anything, do you think the conscious mind does? It seems you answer any and all questions with 'the unconscious mind'. It leaves to to question if the concious mind plays any role at all in your philosophy.

Maybe they should be changed from "a robot" to "any AI"

I'm fairly sure that the idea is that any AI will be a robot, almost by definition. Remember, robots don't have to look like people, that is an android, they only have to have some ability to modify its surroundings, real or virtual. Some examples of real and fictional robots include KITT from Knight Rider, Hal 9000 from 2001: A Space Odyssey, and the toilet that watches you pee so it can flush when you are done, and the paper towel dispenser that rolls out the next piece when you tear off the current one. Never forget, robots are watching you pee.

Why would anyone be concerned about the ethical treatment of something that doesn't exist? This is less important than the make up of pocket lint. Butterfly farts are going to have more effect on the world than AI.

I think the idea is that because our society is so computationally dependant that as soon as the first AI is born it is going to become super powerful, so we better have already devised a way to keep it from hating us.

Personally I think this is pure science fiction. I think that if we do invent an AI it is going to be a moron. It will take many years, and many iterations to make a super powerful AI, if we ever do.

I think we should be much more concerned with our complete inability to treat each other ethically rather than worrying about a maybe scenario about a super powerful computational intelligence. Honestly if we can't treat other people that are 99.99% identical to us with kindness what chance do we ever have of treating something so unlike us with even a modicum of compassion?

 

[Moonbeam]

What, if anything, do you think the conscious mind does? It seems you answer any and all questions with 'the unconscious mind'. It leaves to to question if the concious mind plays any role at all in your philosophy.

Let's say that all my lifeI was taught it was weakness to have any sympathy for the weakness of others, so one time when I was down on my luck and a Good Samaritan wanted to help me I killed him. He told me he was worthless himself and implied that I was weak because I had those unconscious feelings. My conscious mind played only a small part. It drew my gun, aimed, and pulled the trigger, knowing I was justified. He who is unconscious of truth is conscious of delusions.

PS: we may all be genetically nearly identical, but we can be ethically quite different. Let us hope the machines are as interested in how love and feeling for others came into the universe. Let's hope they cas see we are motivated in hope part because we believe, having bees put down as children, that we are worthless things full of rage at being made to feel that.

 

[Greenman]

Watch 50 years from now your quote will popup on nightly news as AI computers start hacking on their own or something else sinister.

Next documentary will be: AI. Where did we go wrong?
The investigator will have with them two psychologists. One being a real human being and one being an AI unit of sorts.

Not even maybe. No arms, no legs, no hands, no way to manipulate the physical world, no hormones, no emotions, no instinct, no drive to succeed. On top of all that, it has to be plugged in, and anything that's plugged in can be unplugged.

 

[inachu]

Not even maybe. No arms, no legs, no hands, no way to manipulate the physical world, no hormones, no emotions, no instinct, no drive to succeed. On top of all that, it has to be plugged in, and anything that's plugged in can be unplugged.

True AI will be sentient. To be self aware. To be self aware will also have the inate ability to save itself or back itself up.

I super smart AI would recompile itself so it can run more smoothly/efficiently.

Our #1 goal is to be its true friend. Just like animals we will need to be its mother/father

The goal and desires will be self driven.

I do think for the first 5 years the AI unit should be sheltered for any outside influences. After 5 years we can drop the veil and let it see reality of all the beauty and ugliness that humanity has all to offer.

 

[Greenman]

True AI will be sentient. To be self aware. To be self aware will also have the inate ability to save itself or back itself up.

I super smart AI would recompile itself so it can run more smoothly/efficiently.

Our #1 goal is to be its true friend. Just like animals we will need to be its mother/father

The goal and desires will be self driven.

I do think for the first 5 years the AI unit should be sheltered for any outside influences. After 5 years we can drop the veil and let it see reality of all the beauty and ugliness that humanity has all to offer.

This is a dream world discussion. You can decide that the thing runs on clean living and pure thoughts, but that doesn't make it possible. What we're actually discussing is how this would be in your fantasy world, there is no give and take because every objection that's made you'll answer with some magical new idea or technology. We'd be as well off discussing all the uses for unicorn horn.

Foolish thread is foolish, I should have never allowed myself to be sucked into it.

 

[Merad]

"True" AI is not likely to exist anytime soon. The only model that we have for a sentient mind is the human brain, and we have at best a very primitive and incomplete understanding of how our own brains work. We'll figure out the moral and ethical ramifications when we're closer to having this exist for real.

 

[inachu]

"True" AI is not likely to exist anytime soon. The only model that we have for a sentient mind is the human brain, and we have at best a very primitive and incomplete understanding of how our own brains work. We'll figure out the moral and ethical ramifications when we're closer to having this exist for real.

We do at the current time have self learning machines but these are the version of caveman IQ robotics and if anything what we have learned from the electronic world is that their speed in learning and knowing is way much faster than the speed that humans learn.

If I live to be 100 then I can see a world filled with AI units with the IQ more than that of a 10 year old.

Data collecting
Data sorting
Data relevant scoring
Acquiring knowledge will be done way faster from multiple sources on the fly.

We could build one today if all the people involved really got together but I think that would take lots of time and money.

 

[Jaepheth]

Everything I program has code designed to kill all humans.

Seemed like the most ethical thing to do.

FOR THE GREATER GOOD.


On a more serious note
I think the first "AI" will arise from uploading human consciousness; so the human's ethics will go with it, no additional coding required.
Because no matter how well your AI is coded; someone's going to argue it's not a true AI until you take something that is undeniably already self-aware and plagiarize it wholesale.

Another way of looking at it is that an AI can't be coded. A true AI would be a natural result of sufficiently complex hardware. As soon as you're coding it, it no longer has free will, and thus isn't a true AI.

 

[PhatoseAlpha]

You coding it doesn't magically strip an AI of free will any more then your parent's genetic code stripped you of free will. Or for that matter, that your parents telling you to eat your vegetables stripped you of free will.

Course, that particular way leads to some unpleasant questions about whether or not you actually have free will, and what free will actually means anyway.

Coding an AI isn't going to be controlling it's thoughts, or dictating them. Even outside the realm of sci-fi, and in the realm of existing technology it doesn't work that way. A coder writing a master level chess program does not play the game for the program and define it in that way. Nor could they reasonably predict it's behavior - the moves it would make.

 

[BigDH01]

Not even maybe. No arms, no legs, no hands, no way to manipulate the physical world, no hormones, no emotions, no instinct, no drive to succeed. On top of all that, it has to be plugged in, and anything that's plugged in can be unplugged.

But it's very existence drastically increases the probability that we are not real. That might have some interesting implications.

 

[SMOGZINN]

But it's very existence drastically increases the probability that we are not real. That might have some interesting implications.

How would we not be real? What definition of real could you have that would not include us?

 

[WHAMPOM]

I thought a guide for making true AI would exist already.

People like Elon Mush and Hawking have made a mad rush for AI to be subservient to us right off the bat without think that if AI is true AI then it must be equal to us in every way.


Anyone know of any legit website dedicated to the ethical guidelines in making true AI?

Define free will as it works in human life and you will have your start in building an AI machine of course people will argue that humans have no free will in the first place.

edit: oh shit I posted in discussion club, shoot me now!