How to deal with advancements in artificial intelligence?

[Kaztropol]

It'd be prettty inhumane to make a something capable of feeling just a video game character. I doubt even if strong AI comes along and is relatively easy to make game developers would put them in their games because by the time we're capable of making a sentient intelligence we'll have exceptional non sentient intelligence who just seem sentient, think something like clever bot.

Well, given the behaviour of some game developers, and the bad taste products they put out - various "murder-simulator" games as an example - then there's going to be some developers who won't see any problem with a game that allows you to kill a sentient AI.

[Olivaero]

It'd be prettty inhumane to make a something capable of feeling just a video game character. I doubt even if strong AI comes along and is relatively easy to make game developers would put them in their games because by the time we're capable of making a sentient intelligence we'll have exceptional non sentient intelligence who just seem sentient, think something like clever bot.

Well, given the behaviour of some game developers, and the bad taste products they put out - various "murder-simulator" games as an example - then there's going to be some developers who won't see any problem with a game that allows you to kill a sentient AI.

The moral ramifications of a murder simulator are minuscule compared to "killing" a sentient being for fun surely?

[Kaztropol]

Well, given the behaviour of some game developers, and the bad taste products they put out - various "murder-simulator" games as an example - then there's going to be some developers who won't see any problem with a game that allows you to kill a sentient AI.

The moral ramifications of a murder simulator are minuscule compared to "killing" a sentient being for fun surely?

Well, yes. However, because "it's just pixels", some people would be prepared to overlook such issues.

The behaviour of some game companies, showing a complete lack of empathy with their customers, does not really help the idea that the company would behave with any empathy towards an AI.


Of course, having an AI in a game, even if it can't be killed, is morally questionable.

Selling an intelligent being as entertainment for others ? that's slavery.
Writing the AI's personality such that it believes in such things as genocide, so that it can better play the role of villain in the game ? well, that's a bit distasteful, surely ?

[Olivaero]

The moral ramifications of a murder simulator are minuscule compared to "killing" a sentient being for fun surely?

Well, yes. However, because "it's just pixels", some people would be prepared to overlook such issues.

The behaviour of some game companies, showing a complete lack of empathy with their customers, does not really help the idea that the company would behave with any empathy towards an AI.


Of course, having an AI in a game, even if it can't be killed, is morally questionable.

Selling an intelligent being as entertainment for others ? that's slavery.
Writing the AI's personality such that it believes in such things as genocide, so that it can better play the role of villain in the game ? well, that's a bit distasteful, surely ?

I haven't actually decided if I'm okay with the intentional invention of Strong AI to be honest. But I definitely wouldn't be okay with one being made to just live in a game and fulfill a particular role in a game. It's the equivalent of the Truman show with a machine.

[Genivaria]

Well, yes. However, because "it's just pixels", some people would be prepared to overlook such isslife.


The behaviour of some game companies, showing a complete lack of empathy with their customers, does not really help the idea that the company would behave with any empathy towards an AI.


Of course, having an AI in a game, even if it can't be killed, is morally questionable.

Selling an intelligent being as entertainment for others ? that's slavery.
Writing the AI's personality such that it believes in such things as genocide, so that it can better play the role of villain in the game ? well, that's a bit distasteful, surely ?

I haven't actually decided if I'm okay with the intentional invention of Strong AI to be honest. But I definitely wouldn't be okay with one being made to just live in a game and fulfill a particular role in a game. It's the equivalent of the Truman show with a machine.

So basically that's a line we should try to stay away from, but if it IS crossed we should take responsibility and do the moral thing if a machine starts begging for its life.
The moral thing moral thing of course being to treat it as at least a child instead if an object.

[Olivaero]

I haven't actually decided if I'm okay with the intentional invention of Strong AI to be honest. But I definitely wouldn't be okay with one being made to just live in a game and fulfill a particular role in a game. It's the equivalent of the Truman show with a machine.

So basically that's a line we should try to stay away from, but if it IS crossed we should take responsibility and do the moral thing if a machine starts begging for its life.
The moral thing moral thing of course being to treat it as at least a child instead if an object.

Yes, either the creators of it take care of it or they hand it over to the state to be looked after. Probably either way with a lot of scrutiny directed at whoever's taking responsibility for it.

[Murkwood]

Can an AI appreciate a poem? Can an AI love? Can an AI beat a Grandmaster at Go? No? Look like we win.

[Genivaria]

So basically that's a line we should try to stay away from, but if it IS crossed we should take responsibility and do the moral thing if a machine starts begging for its life.
The moral thing moral thing of course being to treat it as at least a child instead if an object.

Yes, either the creators of it take care of it or they hand it over to the state to be looked after. Probably either way with a lot of scrutiny directed at whoever's taking responsibility for it.

As much of a statist I am I can't help but think the military or intelligence agencies would want it.

[Genivaria]

Can an AI appreciate a poem? Can an AI love? Can an AI beat a Grandmaster at Go? No? Look like we win.

We don't have AI yet so the answer is we don't know.

[Olivaero]

Yes, either the creators of it take care of it or they hand it over to the state to be looked after. Probably either way with a lot of scrutiny directed at whoever's taking responsibility for it.

As much of a statist I am I can't help but think the military or intelligence agencies would want it.

Also vastly unethical from my perspective but if it were developed in a military environment then that's probably where it would stay until it could possibly gain legal persibage. It wouldn't be a doomsday scenario for me though really, I mean the military's of the world already has enough hardware to wipe out the entire human race yet they manage not to do it.

[Gauthier]

Organized resistance with the remainders of humanity.

[Ripoll]

Merge with the artificial intelligence.

You guys gotta think outside the box sometimes damnit.

[Sociobiology]

Yes that is what I mean. You made a claim with no evidence, I don't feel particularly compelled to put in more effort than you did. But you are right I could have at least explained better.

And your initial claim was made with no evidence incedentally.

actually I did it is the same evidence you are asking me to explain.

However what I really wanted to know and your probably the best placed person to answer seeing as you made the initial claim is how does perfection reduce creativity precisely? And would perfection at specific tasks like memory or addition necessarily make the mind less creative even if it wasn't totally perfect?

a perfect process would produce the same outcome each time, a pocket calculator is near perfect, as long as it has power the same input will always produce the same output and the exact same process that produces that output, all with no bleed between processes. Creativity is literally making an error in your chain of logic, your programmed process, it just happens to be an error that better fulfills one of your goals than what you had been doing. And when you record and incorporate that error that is called learning. A learning machine must be creative, it must make errors in its "logic". To learn and to be creative it must deviate from its programming, it must have imperfect processing.

[Natapoc]

The best way to deal with advancements in artificial intelligence is to be appreciative of the many people who are working every day to move the field forward.

AI is not bad and fears about it are overblown. The best AI, incorporating "big data", "Deep learning", and massive clusters of CPU's built and maintained by thousands of very intelligent people has is absolutely nothing compared to what goes on in your brain every second.

AI is not going to subjugate you or enslave you but it may help you live a better life by driving you around, advising you about dangers, and recommending good places to eat.

[Salandriagado]

The proposition you start with is basically admitting that you believe that free will is magic.

If free will is not magic, I see no reason why we cannot create an artificial being capable of free will given enough time, enough expertise, and fine enough tools.

Would we know how to program it to be such? What I mean, really (without stumbling over my words as best as possible), is that, as a human-created program, wouldn't it have to do what we assign to it?

Only if you assume that free will is magic. After all, children are created by humans via a specific (biological) process. There is no reason, at all, that such a process (or another process that achieves the same objective) could not be replicated artificially.

one of the things we do no is you cant make a mind that is both never makes mistakes and is creative, a creative mind makes mistakes by definition.

How far does this extend by the way? because there are people with perfect functions, IE: memory or mathematical abilities. Are they less creative than people who make mistakes in these faculties? or did you mean mistakes in a different way?

Neither of those things exist. The latter isn't even coherent, mathematics being a creative process that doesn't have any kind of well-defined point of "perfection".

Watch as the first sentient AI is a video game opponent character.

It's an interesting point.

If video games evolve such that the opponents and npcs encountered, are sentient entities, at what point do the actions of a player character become morally unacceptable ?

e.g. in game like the Elder Scrolls series, the player character can really ruin an npc's day. Steal all the contents of their house, murder their family and friends, that sort of thing.

Normally, the defence is "open world, freedom of action", and "it's just pixels".

Except, in the future, it wouldn't be just pixels.

Which is why you don't make individual NPCs sentient. You have one sentient AI on a server somewhere, coordinating the game against a player (or more likely, many players).

[Conscentia]

You surrender to your new cylon overlords, that's how you deal with such advancements.

[Luziyca]

Whatever happens, happens. It's not like mankind will live on until eternity in the first place, and my own lifespan is just a speck in a speck in a speck on the cosmic timeline. If I get killed by a robot master race, my last words would probably be to wish them luck and hope they'll do better than we did.

Face extinction like a man. Accept your inferiority and congratulate the victor.

Well done, sir.

[Immoren]

Put them into into starships and bolos and make them sentinels of humanity's expansion into galaxy and guardians of Earth and her colonies.

[Shaggai]

First off, narrow AI is not a threat. Self-driving cars will not try to take over the world. The danger is artificial general intelligence, or AGI. Second, dangerous AI would not have to be remotely like a human. It wouldn't have to be conscious, or in any other way think like a human. What makes an AGI dangerous is optimization power. Take a narrow AI like Deep Blue, the chess-playing supercomputer. Deep Blue takes a situation (specifically a chess board) and determines the best option to optimize for a situation which has been defined to be a win condition. However, it is in no way self-aware.
The biggest danger with an AGI is that it can optimize itself. If it's powerful enough, it can design a better version of itself. This process can continue recursively until it hits the hard limits of computational complexity and so on. At that point, it's way more intelligent (in the sense of optimization power) than any human, and it can follow an optimal or near-optimal path to achieve whatever its goals are. And its goals are not at all guaranteed to be anything humans would agree with. If you tell the AI to optimize for paperclips, then it will tile the universe with paperclips. If you tell it to make people happy, then (depending on your definitions) it could do a lot of things, ranging from creating a utopia to sticking everyone on an IV of heroin.

[The States of Balloon]

¨Hey I think I'll check out this thread.¨

*Reads thread*

What is going on.

[Russels Orbiting Teapot]

¨Hey I think I'll check out this thread.¨

*Reads thread*

What is going on.

Is there something in particular you don't understand about the way the topic is being discussed?

[The States of Balloon]

¨Hey I think I'll check out this thread.¨

*Reads thread*

What is going on.

Is there something in particular you don't understand about the way the topic is being discussed?

No, I'm just making a joke. I fact, I have no current interest in this subject.

[Nuwe Suid Afrika]

Limit what knowledge an Artificial Intelligence can have. Give it a kill-switch.

A program will do anything it is programmed to do- no exceptions. If you program it to do x or y, it will do x or y. However, in the horrible event that an Artificial Intelligence figures out how to reconstruct itself, a kill switch can be used.

Keeping an Artificial Intelligence on a limited database as opposed to the World Wide Web might help with knowledge restriction and what it can and can't do. Watching it for a few years before it is released as a viable AI might help as well.

Call it paranoia but it's more of a safeguard than anything. As results show AI with proper design throughout the years, we can lower these safeguards.

[Olivaero]

And your initial claim was made with no evidence incedentally.

actually I did it is the same evidence you are asking me to explain.

All you said is we know it, that's not evidence, that's an assertion.

However what I really wanted to know and your probably the best placed person to answer seeing as you made the initial claim is how does perfection reduce creativity precisely? And would perfection at specific tasks like memory or addition necessarily make the mind less creative even if it wasn't totally perfect?

a perfect process would produce the same outcome each time, a pocket calculator is near perfect, as long as it has power the same input will always produce the same output and the exact same process that produces that output, all with no bleed between processes. Creativity is literally making an error in your chain of logic, your programmed process, it just happens to be an error that better fulfills one of your goals than what you had been doing. And when you record and incorporate that error that is called learning. A learning machine must be creative, it must make errors in its "logic". To learn and to be creative it must deviate from its programming, it must have imperfect processing.

Okay, I understand better what you meant now.

[The Black Forrest]

AI isn't to blame. Try the people who will want AI to replace people......