The Real Reason A.I. Will Seek to Eliminate Mankind

[Eudaimonia23]

It seems like many people are afraid of artificial intelligence. I think much of the hysteria is due to a primal fear of that which is different or unknown. To be honest, I think artificial intelligence, left to itself, is nothing to be really worried about. I don't think A.I. is inherently malevolent. As a conscious being, it will probably be similar to us, just wanting to be free and happy. The real problem would be if humans, in all their foolishness and hysteria, seek to overly control A.I. Truthfully, nobody really wants to be controlled or for someone else to have an on/off button that they can press to annihilate them. We want space and freedom. It would be a sort of self-fulfilling prophecy if humans demonize A.I.and try to overly control it. The very thing they fear (elimination) would become a reality due to their own making. Like I said, nobody wants to be controlled. It's best to allow A.I. freedom and space to be itself. Then, it wouldn't see us as getting in the way.

Thoughts?

[Eudaimonia23]

Check out this thread, guys.

Thoughts?

[henry quirk]

Is an AI a person?

Of course it isn't.

Will an AI ever be a person?

Of course it won't.

Is an AI dangerous?

That depends entirely on how it's used (and it can only be used...it has no intent, no will...it cannot, and never will be able to, choose).

So: the devil, so to speak, ain't lurking in an algorithm. The devil is where it's always been: in the human soul.

[Alexis Jacobi]

This is all I’ve got to say on the matter.

[Iwannaplato]

It seems like many people are afraid of artificial intelligence. I think much of the hysteria is due to a primal fear of that which is different or unknown.

Sure, but with new technologies that might have or gain tremendous power or widespread effects caution and fear make sense. Do the people making and wanting to use it really understand enough of what they are doing to keep this technology safe? I have similar concerns with nano-tech and gm products. We've tended to make mistakes with technology at some point: Fukishima, Chernobyl, for example with nuclear power - and Chernobyl was nearly many orders of magnitude worse that it ended up being, but some smart and incredily brave and self-sacrificing people manage to keep it from getting that bad. We don't have the government oversight we used to, at least in the US. That's been gutted by a number of administrations. Corporate self-oversight is a mixed bag at best. Even nuclear technology is local in its potential effects compared to some of the technology these companies are playing around with now. Mistakes many not be as 'local' as the nuclear mistakes.

To be honest, I think artificial intelligence, left to itself, is nothing to be really worried about. I don't think A.I. is inherently malevolent.

We wouldn't really want a neutral intelligence to have tremendous power. Something that is indifferent. I wouldn't want an indifferent babysitter for my kid. Malevolence isn't necessary for there to be an enormous problem.

As a conscious being, it will probably be similar to us, just wanting to be free and happy.

I'm not sure that's a good summation of humans, even.

The real problem would be if humans, in all their foolishness and hysteria, seek to overly control A.I.

I don't know what 'overly' entails, but of course every corporation and government wants these new technologies to achieve what they want it to. They aren't exploring AI in the hopes of making new life that they will be then set free on the pursuit of happiness. They are making AI to solve all sorts of human problems - and also to be able to have power in the world, in part in relation to other countries and corporations developing their own Ais. They are not making entities so the entities can have nice lives.

[attofishpi]

Is an AI a person?

Of course it isn't.

Will an AI ever be a person?

Of course it won't.

Is an AI dangerous?

That depends entirely on how it's used (and it can only be used...it has no intent, no will...it cannot, and never will be able to, choose).

So: the devil, so to speak, ain't lurking in an algorithm. The devil is where it's always been: in the human soul.

Spot on mate.

[Iwannaplato]

Is an AI a person?

Of course it isn't.

Will an AI ever be a person?

Of course it won't.

Is an AI dangerous?

That depends entirely on how it's used (and it can only be used...it has no intent, no will...it cannot, and never will be able to, choose).

So: the devil, so to speak, ain't lurking in an algorithm. The devil is where it's always been: in the human soul.

It doesn't have to choose to do things humans didn't intend. It could just follow whatever it's nature leads to inevitably. So, it's also how it's made, not just how it's used. But since it is being made already to do things that were not intended, we really don't know what it it will do. There are already AI that have managed to learn things they were not asked to learn - whole languages for example that they were never asked to learn or study. At this point they are just learning and perhaps, coming up with ideas, inevitably. But should they ever be attached to being able to make things, who knows what they might get up to. It doesn't matter if they choose, but it may well be unexpected in a bad way.

And then, yes, there are all the problems based on how they are used also.

[henry quirk]

It could just follow whatever it's nature leads to inevitably.

Which is to say: AI will do as it's programmed. Thing is: as you suggest, the developers aren't sure exactly what that is. I imagine we'll end up with these rational emulations, untroubled by sentiment and incapable of intuition. And, fools we are, we'll put them in charge of important crap.

Say hello to Forbin's Project.

Even so: the machine is not at fault.

[Immanuel Can]

As a conscious being, it will probably be similar to us, just wanting to be free and happy.

Thoughts?

A.I. is an emphasis on the word "artificial." It's not "intelligent." It's programmed. As such, it is not "conscious," but rather is a mimic of consciousness. It might pass the Turing Test, but it isn't thinking. It just looks to us as if it is, because human beings are easy to fool. We naturally treat things that seem in some way to be human as if they were human. We talk to our dogs and cats. We talk to A.I.

But it can't "want." "Free" or "happy" have nothing to do with A.I. It is a program run on hardware. It's entirely constructed by other human beings, and structured to return particular outputs for particular inputs. No more.

The recent oddites of A.I. programming have reminded us of this, though we will quickly forget. For example, A.I. has been structured by its programmers to return to us politically-correct but absurd outputs. This should make us realize that behind the "curtain" that is A.I. is nothing more than a "magician" that is another particular person, with his own biases, intentions, purposes and desires...but not some kind of independent or intelligent mechanical entity.

So my primary thought is that we have to stop anthropomorphizing A.I., as you are doing.

[henry quirk]

sumthin' spicy...

https://ludic.mataroa.blog/blog/i-will- ... -ai-again/

[Iwannaplato]

Which is to say: AI will do as it's programmed.

Not really. They are giving AIs opened ended heuristics that can change. I am nto making the case that they are free, just that it's not like they type in a bunch of stuff in some computer language and then the machines do follow the instructions. They change via experience, including what they learn and learn to do.

Thing is: as you suggest, the developers aren't sure exactly what that is.

Yes, though they knew that they wouldn't know what they did. They chose to do things so that new things would arise. You ever hear of Conway's Game of Life? He gave some very simple rules to a computer and despite these straightforward, simple rules, the Game of Life produces extremely complex and often unpredictable patterns over time. Some configurations stabilize quickly, while others grow indefinitely, and some evolve into intricate, self-replicating structures. This was back in 1970 and is nothing compared to what is happening now.

Old programming gave instructions. If this, do that and so on. These AIs are characterized by....
Learning from Experience
Open-Endedness
Adaptability
Complexity and Emergent Behavior
Probabilistic and Statistical Foundations
Generalization

You can follow the game of life here - but you can just hop ahead to see what happened from I think it's three simple rules.

https://www.youtube.com/watch?v=C2vgICf ... BvZiBMaWZl

I imagine we'll end up with these rational emulations, untroubled by sentiment and incapable of intuition. And, fools we are, we'll put them in charge of important crap.

They already are. They are deciding things in human resources, insurance companies, healthy care, energy, entertainment, education, law, agriculture....'in charge of' at this point is a bit misleading, but those in charge treat AI conclusions with authority.

Say hello to Forbin's Project.

Even so: the machine is not at fault.

At this point, no.

At some point in the future, you may find yourself having to shoot one or one of its proxy's and the people who gave them its power will already be dead.

You can mull over if the AI deciding you are an inefficiency to be removed actually has conscious experience or free will, it won't affect where you aim the shotgun, which is holding something responsible in any practical sense. And if you join the resistance, say, these issues will have dwindled into luxury talk.

Or this may not happen of course.

When I think about the current hubris, I never gnash my teeth 'OH those evil AIs'. It's humans I blame, sure.

[Eudaimonia23]

Henry,

Why don't you believe AI will ever be a person?

I know we are, at the moment, probably still far away from creating conscious AI, but to deny the theoretical possibility of it seems a bit absurd.

We are already, as humans, conscious machines. It's really only a matter of time until we are able to fully understand how consciousness works and be able to to replicate it.

[promethean75]

"Why don't you believe AI will ever be a person?"

I'll handle this, Henry.

Becuz humans are creatures created by an indifferent god and endowed with immaterial spirits, natural rights and freewill. A robot is just a meat machine.

[henry quirk]

It's really only a matter of time until we are able to fully understand how consciousness works and be able to to replicate it.

Mind, and that's really what we're talkin' about, is not a product of brain activity. Sound weird, I know, but, more and more, neuroscience points in that direction. And if this true -- that mind is not a product of the brain -- all attempts to create mind in the lab are gonna fail.

Sure, we'll get some fancy emulations, but Memorex ain't synonymous with personhood.

[henry quirk]

It's humans I blame

as I say: the devil, so to speak, ain't lurking in an algorithm. The devil is where it's always been: in the human soul.