[ARCHIVED THREAD] - Why not train AI to kill? (Page 1 of 2)
Posted: 12/3/2016 8:52:19 PM EDT
| [Removed and account locked - Paul] |
| Come on now. Skynet is a character and a concept in a movie. This is real life. Such a thing will not be possible. Machines will never become 'self aware' as we are. They can merely be programmed to give the appearance of self awareness but only for the purpose of being usable by humans. |
|
Quoted:
Come on now. Skynet is a character and a concept in a movie. This is real life. Such a thing will not be possible. Machines will never become 'self aware' as we are. They can merely be programmed to give the appearance of self awareness but only for the purpose of being usable by humans. I'll take the dozens of scientist's warnings over yours. 
|
|
Quoted:
Deep learning. Why not train some AI by watching a shit ton of first person video footage of people in combat. Telling it whats a good idea, whats not a good idea, what is a threat and what isn't. It should be easy to tell a machine that a human holding a gun, or holding a gun at an certain angle (pointing it at the machine) is a threat and what to do. Why risk human lives in combat when we can send machines? What enemy would threaten us if we had nothing to risk but machines? lol you're either trolling or. . . . . 
|
|
Quoted:
Come on now. Skynet is a character and a concept in a movie. This is real life. Such a thing will not be possible. Machines will never become 'self aware' as we are. They can merely be programmed to give the appearance of self awareness but only for the purpose of being usable by humans. BULLSHIT It's a how to guide. |
|
Quoted:
Come on now. Skynet is a character and a concept in a movie. This is real life. Such a thing will not be possible. Machines will never become 'self aware' as we are. They can merely be programmed to give the appearance of self awareness but only for the purpose of being usable by humans. Confirmed OP is a bot sent to seek out public opinion before the war |
|
Quoted:
Come on now. Skynet is a character and a concept in a movie. This is real life. Such a thing will not be possible. Machines will never become 'self aware' as we are. They can merely be programmed to give the appearance of self awareness but only for the purpose of being usable by humans. That's just what skynet what's you to think. |
|
I have no doubt that this will happen, sooner or later. I wouldn't be even surprised if several countries already have plans to implement it if another country does.
Even if the AI were not "self-aware," it could get really ugly. When it does become self-aware, we're done. |
|
Quoted:
Ok smartasses. Tell me what you think when your grandkids are the ones being killed instead of a machine. I promise you we can make killer robots that will not turn on us. Unless we programmed them to and didn't include some sort of kill switch. This was fun. Same time tomorrow night? |
|
Quoted:
Ok smartasses. Tell me what you think when your grandkids are the ones being killed instead of a machine. I promise you we can make killer robots that will not turn on us. Unless we programmed them to and didn't include some sort of kill switch. First, how about a phone that doesn't self-destruct? Get that one right, then come see me about killer-AI. |
|
Quoted:
Umm, hacking??? Also the Three Laws of Asimov would prevent this...  This. Say someone were to upload a virus into your AI army while it is state side. You just solved china's problem of getting their armies over the seas. It seems like Pandora's box to me. |
|
Quoted:
I have no doubt that this will happen, sooner or later. I wouldn't be even surprised if several countries already have plans to implement it if another country does. Even if the AI were not "self-aware," it could get really ugly. When it does become self-aware, we're done. We don't need to worry about it becoming self aware. And what I'm thinking about would be like a weapon not some sort of amorphous software like skynet was that could just "install itself" on everything and then magically take over ICBMs and launch them. That is FANTASY. Its not possible in reality and in order for it to be we would need to specifically change a whole lot of things in order for it to be possible. This is like a gun with legs or whatever form of locomotion it takes on (by our design) that can figure out what to shoot at and shoot itself. The possibility of it shooting a friendly robot or person? Possible, in theory. The possibility of it somehow developing grand tactical schemes and plans for the future and overthrowing humanity? Non existent. It only knows what we train it to know based on the information we give it. And that information would be limited to what we provide it, it would know how to move and shoot things that are moving based on what it sees visually. It would know how to travel down a road or how to avoid obstacles, open doors, clear a building, things like that. It would know if someone is holding their hands up that they're surrendering and not to shoot them. The enemy might figure that out and use it against them, and we would have to improve that training. |
|
Quoted:
We don't need to worry about it becoming self aware. And what I'm thinking about would be like a weapon not some sort of amorphous software like skynet was that could just "install itself" on everything and then magically take over ICBMs and launch them. That is FANTASY. Its not possible in reality and in order for it to be we would need to specifically change a whole lot of things in order for it to be possible. This is like a gun with legs or whatever form of locomotion it takes on (by our design) that can figure out what to shoot at and shoot itself. The possibility of it shooting a friendly robot or person? Possible, in theory. The possibility of it somehow developing grand tactical schemes and plans for the future and overthrowing humanity? Non existent. It only knows what we train it to know based on the information we give it. And that information would be limited to what we provide it, it would know how to move and shoot things that are moving based on what it sees visually. It would know how to travel down a road or how to avoid obstacles, open doors, clear a building, things like that. It would know if someone is holding their hands up that they're surrendering and not to shoot them. The enemy might figure that out and use it against them, and we would have to improve that training. GD is a rough place. I don't think you'll find what you're looking for here. |
|
OP isn't trolling so I need to actually tell him.
I have a hacker friend He can shut down pretty much any connection Robot no work after that He could easily make a program to make nonviolent things be seen as acts of aggression and trigger a self defense response But even if total cyber security was acheived( practically impossible) 1 EMP could shut down all of it 1 solar flare, one power spike, one low yield nuclear device..... Robots are super delicate They would break way easier than a person They are simply not suitable to take over all military functions. One hacker could install a backdoor That means that person could control every damn robot that used that software When robots finally are strong enough and resilient enough for combat, then sentience actually could happen Nobody wants that OP Think of the children |
|
Quoted:
Come on now. Skynet is a character and a concept in a movie. This is real life. Such a thing will not be possible. Machines will never become 'self aware' as we are. They can merely be programmed to give the appearance of self awareness but only for the purpose of being usable by humans. This. Every decision a computer makes is the result of programming. For true AI to work, computers/software will have to replicate complex neural networks similar to our brains. So far, at least for now and the foreseeable future, only nature can create neural networks with any measure of success. It would require a technology level jump of a massive scale to move us into true AI. Maybe in the future but I doubt it will happen in my lifetime. |
|
Quoted:
Ok smartasses. Tell me what you think when your grandkids are the ones being killed instead of a machine. I promise you we can make killer robots that will not turn on us. Unless we programmed them to and didn't include some sort of kill switch. Who is this "we"? |
|
Quoted:
Come on now. Skynet is a character and a concept in a movie. This is real life. Such a thing will not be possible. Machines will never become 'self aware' as we are. They can merely be programmed to give the appearance of self awareness but only for the purpose of being usable by humans. Not true |