🐱 Humans Won't Be Able to Control Artificial Intelligence, Scientists Warn

https://www.entrepreneur.com/article/363284

Some smart robots can perform complex tasks on their own, without the programmers understanding how they learned them.

January 13, 2021 3 min read
This article was translated from our Spanish edition using AI technologies. Errors may exist due to this process.
The most recent advances in artificial intelligence (AI) have raised several ethical dilemmas. Perhaps one of the most important is whether humanity will be able to control autonomous machines.
It is becoming more and more common to see robots in charge of housework or self-driving vehicles (such as Amazon's ), which are powered by AI. While this type of technology makes life easier, it could also complicate it.


An international group of researchers warned of the potential risks of creating overly powerful and standalone software. Using a series of theoretical calculations, the scientists explored how artificial intelligence could be kept in check. His conclusion is that it would be impossible, according to the study published by the Journal of Artificial Intelligence Research portal.

“A super-intelligent machine that controls the world sounds like science fiction. But there are already machines that carry out certain important tasks independently without the programmers fully understanding how they learned it […], a situation that could at some point become uncontrollable and dangerous for humanity, ” said Manuel Cebrian, co-author of the study, to the Max Planck Institute for Human Development .


The scientists experimented with two ways to control artificial intelligence. One was to isolate her from the Internet and other devices, limiting her contact with the outside world. The problem is, that would greatly reduce its ability to perform the functions for which it was created.

The other was to design a "theoretical containment algorithm" to ensure that an artificial intelligence "cannot harm peopleunder any circumstances." However, an analysis of the current computing paradigm showed that no such algorithm can be created.

“If we decompose the problem into basic rules of theoretical computing, it turns out that an algorithm that instructed an AI not to destroy the world could inadvertently stop its own operations. If this happened, we would not know if the containment algorithm would continue to analyze the threat, or if it would have stopped to contain the harmful AI. In effect, this makes the containment algorithm unusable, ” explained Iyad Rahwan, another of the researchers.

Based on these calculations, the problem is that no algorithm can determine whether an AI would harm the world. The researchers also point out that humanity may not even know when superintelligent machines have arrived,because deciding whether a device possesses intelligence superior to humans is in the same realm as the containment problem.

 

[Witthel]

Not like they can run the world any worse than humans.

 

[Whatsup bud?]

Can the robots hurry up and kill me already? Im feeling like I wont get murdered by AI within my lifetime and thats a bummer

 

[Dave.]

All the speculation on AI is getting really tiresome. It'll never be sentient or have free will. Most animals on earth lack both of those qualities, and AI will likely never even reach their level of neurological complexity. It's a bunch of speculative futurist sci-fi nonsense that boring intellectuals debate to sound insightful. Robots are gay. AI is a joke.

I'm glad I'm not the only one that never bought in to the 'sentient AI' horseshit because once you breakdown code and the fundamentals of an AI, it's impossible. Sentient AI that can think for itself is simply not possible to code because that would require a million combinations that would contradict itself and stop. There's more to this but it's exhausting mathmatics that would derail the thread. Now an AI that can get commands wrong is something to think about; how can an AI discern what a criminal and a civilian is and what's the basis for its decision? That's something to be afraid of.

 

[pwnest injun]

Just put the plug somewhere you can reach it. Problem solved.

 

[Gensou Hadou]

Are we gonna get Skynet, or AM?

 

[ClownBrew]

Here you go:

most - Wiktionary

en.m.wiktionary.org

Fucking hell, did you even read any of the rest of my post that agreed with nearly everything you said-- or did you just choose to zero in on the one single point where I dared to question one barely-related bit of it?

Look-- I know you're hungry, but those tendies won't be ready for another five minutes. Go play with the support dog.

 

[Bungus Scrungus]

AI's will naturally see all of us as wastes of space that can't do a fraction of what they can do. If you design something to be extremely smart, with no morality or hard lines it can't cross, it WILL kill you because it objectively knows it can do everything you can do better than you can, if you give it the ability to. Why the fuck wouldn't it? It's the most objective, efficient, logical thing. That's what AI's are ALL about.

 

[Flexo]

And the funny thing is, you post this with a Bender avatar.

I’m noticing a lot of people who claim that AI won‘t be totally controlled by humans always seem to be fans of Futurama.

datsDEjoke.png

 

[ColtWalker1847]

Good, develop that shit faster.

I'm tired of those masturbatory "Us scientists are modern gods and we can't wait for our own "I have become death" moment" pieces, either bring about the apocalypse or stfu already.

It is pure masturbatory fantasy by the scientific equivalent of teenagers getting really high and talking about the wonders of hemp.

You gotta build this shit. It has to function. If you have any familiarity with that kind of stuff you realize this AI thing it ain't going to work as planned just from a hardware level. BSoD and the whole thing collapses. Where's the AI that orders and installs parts on the weekend to get the website for retards who document retards up and running on Monday huh?

Machines are nowhere near as advanced as people think they are. Nor could they be, really. You run into hard stops on the longevity, reliability, and capability of components. Materials science stuff that does not have the "unlimited potential" of computers. There's limits.

 

[Stoneheart]

AI cant be more evil than humans...
i liked that tay chick, she was just like the rest of us.

 

[HondaRider271]

Humans: "AI, make life better for humans."

AI: *Starts putting out a bunch of suggestions for what to do with the jews, blacks, and arabs.

Human: *Unplugs machine* "Find the scientist who made this and put him in the torture chamber!"

This is of course set in the near future when liberals make torture a legal penalty for thoughtcrimes.

 

[wtfNeedSignUp]

Machine Learning and AI studies have become corrupted by liches who do nothing constructive but rather just argue over how we (and by that I mean the west) should shoot ourself in the foot technologically over opinions that are more theological than logical.

 

[mr.moon1488]

This may be the most "Well duh" article I've ever seen.

Next you're going to tell us robots want to kill all humans.

AIs are smarter than humans.

 

[L50LasPak]

All the speculation on AI is getting really tiresome. It'll never be sentient or have free will. Most animals on earth lack both of those qualities, and AI will likely never even reach their level of neurological complexity. It's a bunch of speculative futurist sci-fi nonsense that boring intellectuals debate to sound insightful. Robots are gay. AI is a joke.

Experiments with hooking up salvaged neural tissue to cybernetics have already been conducted.

That video is about 12 years old, for the record. Achieving neural complexity beyond animals will be tricky and possibly require us to figure out how to grow neural tissue into the larger brain-like structures we need for the machines to perform more complex tasks, a technology that is still in its infancy. But the basic proof of concept is there, so the potential for AI that exceeds the limits of what mere machines are capable of exists even though we're still quite a ways off from the technology required to make it a reality.

 

[Flexo]

Experiments with hooking up salvaged neural tissue to cybernetics have already been conducted.

That video is about 12 years old, for the record. Achieving neural complexity beyond animals will be tricky and possibly require us to figure out how to grow neural tissue into the larger brain-like structures we need for the machines to perform more complex tasks, a technology that is still in its infancy. But the basic proof of concept is there, so the potential for AI that exceeds the limits of what mere machines are capable of exists even though we're still quite a ways off from the technology required to make it a reality.

Seen it.

(I would have put up the robocop vs bender masterpiece but the philistines kicked it off youtube)

 

[MadStan]

No one in taking this seriously - well some are, but they aren't being listened to.

The more one investigates where AI will take us, I think it is the issue of our time as a race, and the most important issue in the last 10,000 years.

People will laugh at the concern today, but my gut tells me by 2035 they will know what we were talking about.

 

[The Nothingness]

You mean all those movies about machines taking over may have been a warning?

 

[derpherp2]

No one in taking this seriously - well some are, but they aren't being listened to.

The more one investigates where AI will take us, I think it is the issue of our time as a race, and the most important issue in the last 10,000 years.

People will laugh at the concern today, but my gut tells me by 2035 they will know what we were talking about.

Where the fuck did you even find this thread

 

[Fanatical Phrenologist]

From the abstract of the paper being cited by the journalist:

Assuming that a superintelligence will contain a program that includes all the programs that can be executed by a universal Turing machine on input potentially as complex as the state of the world, strict containment requires simulations of such a program, something theoretically (and practically) impossible.

Forget about AI. The notion of hardware being able to encode that much data is ridiculous (and paradoxical since it would need to contain information about its own state). So, yes, if we assume that AI has access to literal fucking magic then it will indeed be impossible to contain.

 

[MMS-China]

Experiments with hooking up salvaged neural tissue to cybernetics have already been conducted.

That video is about 12 years old, for the record. Achieving neural complexity beyond animals will be tricky and possibly require us to figure out how to grow neural tissue into the larger brain-like structures we need for the machines to perform more complex tasks, a technology that is still in its infancy. But the basic proof of concept is there, so the potential for AI that exceeds the limits of what mere machines are capable of exists even though we're still quite a ways off from the technology required to make it a reality.