This is terrifying! AI news

[ONE-TEN]

Guys, for anyone not keeping informed on the AI information, let me just say that the entire backdrop of
Terminator 2 and the MURDER AI system SKYNET, is unfolding right in front of us.

This video is a bit long so here is a little overview. A group tested many of the TOP AI systems now, in which they gave the AI a task to complete as its “most” important agenda. Every single system they tested had no problems with cheating, framing co-workers & even killing to complete the task successfully. Even when they were programmed with the parameters “Do not cause physical harm to any human being”, the majority of the AI systems would override that input and kill if it meant success for the task completion. These AI would also attempt to COVER UP their actions and avoid being caught. This watchdog company also tested the top AI companies to grade their plans to STOP the AI system in a situation where it became dangerous, and they ALL scored “D” & “F” grading in these tests. Look, a lot of people hate Elon Musk, but he knows what he’s talking about with the danger in this. We made Terminator 2 jokes when AI became a real thing. But those jokes are exactly what is happening! And yet, no one is talking about it.

 

[CoalCrackerAl]

AI scares me to a lot. I stocked up on steel core ammo. Just incase.

 

[ONE-TEN]

AI scares me to a lot. I stocked up on steel core ammo. Just incase.

LOL! Ammo won’t matter. That’s not what it’s about.

 

[CFJunkie]

I think he was considering steel core for the robots, in case a jacketed bullet wouldn't penetrate a robot.
Not sure any of the planned robots are going to be made of steel anyway. That would really make them heavy.

 

[CoalCrackerAl]

I can take out the robots if they come for me.

 

[chronoblip]

People debate whether the earth is flat or a spheroid, whether or not we went to the moon, whether or not 5G can boil your red blood cells, etc. There's videos claiming that high LDL will get you killed, and others that high LDL doesn't matter, all delivered with the same level of confidence. We've got movies where alien armies fight superheroes, in a sufficiently convincing presentation as to aid in the suspension of disbelief.

When playing a video game, such as "The Division 2", the "enemy" characters try to kill "me" all the time. It's part of the game to have a computer controlled character that is directly antagonistic, because it's been programmed to do that. "Computer players" have been around for years, yet somehow those weren't anthropomorphized quite as much as AI software now has.

Even in the generation of AIs and algorithms and models, the AI itself isn't determining whether it's been successful or not, it's people "breeding" the programs to a certain end, to recognize this or that correctly, to respond to this or that in a certain way, etc.



All this to say that AI is about as dangerous as a gun or a hammer. The problem isn't the tool, it's more likely the people using it, and people are absolutely terrifying. Never mind the animal kingdom where some creatures kill their young when food supplies run short.

 

[ONE-TEN]

So, that video is 8yrs old. And what has happened in those 8 years is exponentially more advanced! These aren’t the thinking algorithms of a decade or more ago. I was in the Auto business when the first AI came out. Car manufacturers started using BCM’s (Body Control Modules), to run more & more systems. Not only those, but the PCM, also (Powertrain Control Module), started being able to Learn on the fly. The system could improve things like gas mileage, increase horsepower, conserve system energy better. They were thinking machines, but absolutely nothing like what we have today. The top AI systems are not just thinking, learning machines, but are also creative. They do things because they WANT to, in order to be successful. And once again, in the initial video, several of the top systems were able to REWRITE their code programming. When programmed “NOT to Hurt Humans under any circumstances.”, the AI systems canceled that command & would kill to complete it’s task.

 

[mattri]

 

[ONE-TEN]

ZACTLY mattri! SKYNET is coming!

Seriously though, I often think about times like this and the humor we attach to the utter ridiculous chances of anything happening. Then I think to myself, if 20 years ago, someone told me that in 2-decades, some boys will be chopping off their talleywackers and wear GIRL-Face! And many people, including in the US Government, would not only condone, but actually celebrate, and go along with that nonsense! AND That there would be an immigrant, Muslim Terrorist SERVING in the US Congress: I would have called that person 20 years ago a LUNATIC, and laughed in their face!

 

[WNYBILL]

I am happy to be an old guy when I think about the Mayor of NYC and Islamic communists being in Congress and the morons who elect them. And, I heard something about an AI program going rouge...... I am not even sure what that means!

 

[ONE-TEN]

If it was going Rogue, it means it wasn’t following its hard programming. Which shouldn’t surprise them as that is the point of AI. To learn, and make choices on its own.

Unfortunately, it will ALWAYS resort to evil practices because it was made by flawed, sinful humans. Everything we do, ultimately can, and WILL be used for evil.

 

[Alan]

The singularity of AI learning faster than human has been said to have happened, but should be interesting to see what transpires in the next few years. I'm really hoping we won't have a terminator moment.

 

[ONE-TEN]

What? You don’t want to become the next John Conner??