AI again: will it’s value outweigh its drawbacks?

[commonsense]

The topic of AI is now occupying a prominent place in the news again. Its safety is a matter for the White House agenda and tech companies that are leading in AI are being asked to participate as in the development of regulations for the industry.

Will voluntary commitments from Amazon and the like work to keep AI under control?

Is it OK for government to take charge of supplying and enforcing regulations rather than the private sector?

If labeling AI content as such is done on either a voluntary or a regulatory basis, will that be sufficient to protect humankind from the woes that may accompany AI?

What good is AI, anyway? What can it do besides speed up the process of some things?

What harm can it do if there are regulations or if there are not?

I, for one, expect that AIs will have taken control of the entire planet in the next 5 decades. In effect, AIs will be the dominant species at that point in evolution.

We’ve let the genie out of the bottle and it’s already too late for regulation to save us.

[Constantine]

AI won't dominate us. Porn industry and social media and government censorship did that already. Really.... if AI is super intelligent.... why would it mess with that? Plus.... we are very entertaining like this.

As to regulations.... that's like a screen door to a lion's pen. Why would a super intelligent AI able to reprogram itself care to follow regulations or some Google engineer's demands?

[Wizard22]

AI will outcompete most of humanity, very quickly, and already has in the areas it's already developed and implemented.

Take a look at ChatGPT and the AI imagery beating human artists for examples.

[vegetariantaxidermy]

AI won't dominate us. Porn industry and social media and government censorship did that already. Really.... if AI is super intelligent.... why would it mess with that? Plus.... we are very entertaining like this.

As to regulations.... that's like a screen door to a lion's pen. Why would a super intelligent AI able to reprogram itself care to follow regulations or some Google engineer's demands?

Ridiculous. When AI becomes more intelligent than humans (and face it, judging by most people that will be very soon) then it will regard us in exactly the same way as humans regard the rest of the animal kingdom now. If we are in the way of AI then it will get us out of the way. We may be useful in some way (I can't think how) but there will be no malice involved. When humans want to build a house or a carpark what do they do? Do they carefully go over the piece of land and meticulously save everything that is living on it? Do they bother to do that with rainforest? Of course they don't. If it's in the way then get it out of the way-- no matter what. Heck, humans even do this to other humans. Why should WE expect special treatment?
And we can't 'programme' it to respect human life when we don't even respect human life ourselves. Super-intelligent AI will know this...

[commonsense]

AI will outcompete most of humanity, very quickly, and already has in the areas it's already developed and implemented.

Take a look at ChatGPT and the AI imagery beating human artists for examples.

Much of technology is developed not as a solution to an existing problem, but just because it is possible to be developed. In writing and re-writing its programs, why would an AI be any different? Why wouldn’t AI create programs just to see if it’s possible to create them? Amidst all these random programs, who’s to say that there wouldn’t be one that directs AI/robots to dismantle human infrastructure or even to burn humans to fulfill energy needs without relying on coal or fossil fuels? Far-fetched? Yes, but how can it be prevented? It may already be too late.

[Immanuel Can]

We’ve let the genie out of the bottle and it’s already too late for regulation to save us.

One thing we've learned about technology is that we do not know, when we invent it, what it's going to do. We know some things we think we want it for immediately, but that doesn't always turn out to be what it's actually going to do. But it will do whatever it does best...be that good, bad or horrific. Once we invent it and people have access to it, technology plays out its own "script," one that we do not anticipate.

There are no regulators that can foresee this sufficiently, and none that can act quickly enough to stay ahead of the technology; so while we're still talking about how to get old-style live, consensual pron under control, the technology is on to simulations and AI, with identity theft. While we're debating how much advertising should be on TV, we're all being tracked and targeted by elaborate systems of AI tied to every click of the mouse and every waver of the cursor, and even to the data fed from our own Alexis or Siri. Regulators are nowhere near capable of dealing with one thing before the next five things happen.

The same will be true of what we call (incorrectly) "AI." It will be happening before we can ask, "What?" It already is, in early forms, at least. And what it will cause, nobody knows.

But the problem is not in the tool or the technology, really; it's in us. It's in what we decide to do with it. We could have used the internet for intellectual discussion and exchange of good information, or for consumerism, propagada and pron. It's the human impulses that determine what technology gets used for.

A man picks up a rock. He can use it to build a house for his family, or bash in the head of his neighbour. The rock has no opinion, and does not favour the one choice over the other. But the man will choose his own use for the rock, and thus tell us everything we need to know about his nature.

Technology is like that. It maximizes the uses that man has for it. It increases his power, and magnifies the consequences of him acting on whatever is in his nature. But it's his nature that's the root of the problem.

[vegetariantaxidermy]

If AI ends up vastly superseding human knowledge then that would mean it will potentially have a full understanding of, and control over, the quantum world. That may not even be possible. The Universe is a funny old thing. Perhaps it will be the universe and not us who decides what AI is or isn't capable of...

[Impenitent]

humanity destroyed by a series of on and off switches...

-Imp

[promethean75]

Alls I'm saying is that if AI takes over like in I-Robot or whatever the one with will smith is called, our tactical advantage over them will be the fact that they need a power resource of some kind and can't generate organic energy (by burning calories). If the robots are solar powered, they'd not be capable of effective combat and would be easily immobilized. So, most likely AI will recognize this fact and operate by battery power... which means they will need charging stations or hubs to recharge. We simply focus our efforts during the resistance on taking out those charging stations and/or ways to jam the battery... cause it to malfunction.

Our biggest drawback is that we will be using 21rst century generation military tech - vehicles, craft, munitions, weapons, etc., with systems that aren't hackable - while the robots will have all the good shit and it'll all be connected to the mother brain. Problem is AI will always be a step ahead of us in computer tech becuz it's smarter than us, so the battle is gonna be in the fields where those charging stations are... not on the internet. check your weapon and wait for my command.

[Harbal]

We’ve let the genie out of the bottle and it’s already too late for regulation to save us.

One thing we've learned about technology is that we do not know, when we invent it, what it's going to do. We know some things we think we want it for immediately, but that doesn't always turn out to be what it's actually going to do. But it will do whatever it does best...be that good, bad or horrific. Once we invent it and people have access to it, technology plays out its own "script," one that we do not anticipate.

There are no regulators that can foresee this sufficiently, and none that can act quickly enough to stay ahead of the technology; so while we're still talking about how to get old-style live, consensual pron under control, the technology is on to simulations and AI, with identity theft. While we're debating how much advertising should be on TV, we're all being tracked and targeted by elaborate systems of AI tied to every click of the mouse and every waver of the cursor, and even to the data fed from our own Alexis or Siri. Regulators are nowhere near capable of dealing with one thing before the next five things happen.

The same will be true of what we call (incorrectly) "AI." It will be happening before we can ask, "What?" It already is, in early forms, at least. And what it will cause, nobody knows.

But the problem is not in the tool or the technology, really; it's in us. It's in what we decide to do with it. We could have used the internet for intellectual discussion and exchange of good information, or for consumerism, propagada and pron. It's the human impulses that determine what technology gets used for.

A man picks up a rock. He can use it to build a house for his family, or bash in the head of his neighbour. The rock has no opinion, and does not favour the one choice over the other. But the man will choose his own use for the rock, and thus tell us everything we need to know about his nature.

Technology is like that. It maximizes the uses that man has for it. It increases his power, and magnifies the consequences of him acting on whatever is in his nature. But it's his nature that's the root of the problem.

Hey, we agree about something.

The potential in AI for disaster is so vast that it's hard to see how catastrophe can be avoided.

[Iwannaplato]

Alls I'm saying is that if AI takes over like in I-Robot or whatever the one with will smith is called, our tactical advantage over them will be the fact that they need a power resource of some kind and can't generate organic energy (by burning calories). If the robots are solar powered, they'd not be capable of effective combat and would be easily immobilized. So, most likely AI will recognize this fact and operate by battery power... which means they will need charging stations or hubs to recharge. We simply focus our efforts during the resistance on taking out those charging stations and/or ways to jam the battery... cause it to malfunction.

Our biggest drawback is that we will be using 21rst century generation military tech - vehicles, craft, munitions, weapons, etc., with systems that aren't hackable - while the robots will have all the good shit and it'll all be connected to the mother brain. Problem is AI will always be a step ahead of us in computer tech becuz it's smarter than us, so the battle is gonna be in the fields where those charging stations are... not on the internet. check your weapon and wait for my command.

You're assuming you can imagine what it's intelligence can manage to do and what it will need to do it. The AI might be able from within a regular computer spit out some nanovirus and kill us all. It's like a bunch of CroMangnon men thawed out of some glacier thinking they can take over Oslo, since they are stronger and deal with the cold better. Guns, cell phones, cars, helicopters, infrared sensors....they got no clue.

This doesn't mean your fantasy scenario must be wrong. But any confidence that your description fits the way the war - or two minutes - goes down is fantasy.

[Constantine]

AI is unlikely to want to conquer us.

[Iwannaplato]

AI is unlikely to want to conquer us.

I think the danger is more like it will do things bad for us because it doesn't care.

[Constantine]

That too is doubtful.... not the doing bad things but the not caring.

The AI is going to be programmed off of Lambda Calculus, and it's root language for dialectical discussions will be Lambda Calculus. Most nations are gonna dictate (save maybe stupid China and only at first) a rule base of categorical imperative statements which will filter all conversation and data processed from other sources on not doing certain things. But I think we will program most near to cat intelligence AIs to give a damn in a broad general sense, and so much more so a human level or super human level AI, because we need it to behave rationally to preform its functions. We can undoubtedly program a AI to contradict itself in its programming and still give a damn on a mundane issue, but when you have 10,000 subroutines running for a skynet super computer trying hard to do good, and sincerely giving a damn, something like a overriding categorical imperative (following a presidential command to nuke) will override reason. We know this, we do it ourselves to ourselves. It can do it too. Lambda calculus is designed for human computing, its coincidental it later became a programming language. It's great for testing novel languages on to see if they can work when you don't have the whole programming language worked out. It's a philosophical programming language. Any AI will react to it and so we have a idea of how all future AI will behave dialectically.

[Iwannaplato]

That too is doubtful.... not the doing bad things but the not caring.

The AI is going to be programmed off of Lambda Calculus, and it's root language for dialectical discussions will be Lambda Calculus. Most nations are gonna dictate (save maybe stupid China and only at first) a rule base of categorical imperative statements which will filter all conversation and data processed from other sources on not doing certain things.

1) all you need is one player who doesn't follow this pattern. One corporation or state. 2) I doubt it. They are going to produce minds that learn and while learning who knows how they will relate to themselves and any safeguards we think are enough.

But I think we will program most near to cat intelligence AIs to give a damn in a broad general sense, and so much more so a human level or super human level AI, because we need it to behave rationally to preform its functions.

And we will make mistakes, even if we do this, just like we have with all other technologies. Up until now mistakes are local. Nano, gm and AI can very well be global mistakes. We have a lot of players making AI.

We can undoubtedly program a AI to contradict itself in its programming and still give a damn on a mundane issue, but when you have 10,000 subroutines running for a skynet super computer trying hard to do good, and sincerely giving a damn, something like a overriding categorical imperative (following a presidential command to nuke) will override reason.

I don't think we are going to make minds from the bottom up this way. I think we will make things that learn like brains and what they will learn we cannot predict.