[ARCHIVED THREAD] - Why do people believe that Artificial Intelligence will have empathy for mankind? (Page 3 of 5)
|
Quoted:
Anyone assuming AI will be benevolent, or will "just do what it's programmed to do", needs to be slapped upside the head. Off the top of my head, I can't think of a more egregious example of wishful thinking. From the dawn of even the concept of AI, the dispassionate reality of machine logic has been the stuff of dystopian science fiction. But, when the reality of AI is developing computers that can help make basic decisions, it's laughable to even suggest the developers should be worried about it taking over their networks and acting like HAL 9000. |
|
Quoted:
Quite easily. https://qz.com/968770/a-robot-reporter-in-china-took-questions-from-wireds-kevin-kelly-and-it-was-super-awkward/ Anyone who thinks the programmers developing this thing actually should be considering the possibility of her becoming Skynet has a few screws loose. |
|
Quoted:
Not a reproduction need but if you could clone ten of yourself to do your work and do it better wouldn't you. Quoted:
Quoted:
You assume that AI will have a reproductive urge... |
|
Quoted:
True no one knows what will happen. I'm just curious and had to ask. Quoted:
Quoted:
Man fears what it does not understand, and you don't understand AI. |
|
Quoted:
I have never heard of anyone suggesting AI" would be "benevolent," "have empathy," or any of these other nonsense terms being tossed about in this thread with zero citation. From the dawn of even the concept of AI, the dispassionate reality of machine logic has been the stuff of dystopian science fiction. But, when the reality of AI is developing computers that can help make basic decisions, it's laughable to even suggest the developers should be worried about it taking over their networks and acting like HAL 9000. |
|
Quoted:
And your point is? Just because man didn't understand and feared fire doesn't mean it wouldn't burn him. Quoted:
Quoted:
The fact remains that you do not understand but do fear AI. |
|
Quoted:
But you're worried about a photograph of a fire burning you. You have a fundamental misunderstanding of the applications and research focus of AI, yet you are all but demanding it be curtailed because, demons might steal your soul. |
You will get shit posted/doxed/then your smart home products killing you at the same time.
|
Quoted:
AI has no imagination. They cannot innovate. They can improve upon existing tech, but cannot create new designs. |
|
Quoted:
Once AI is capable of producing better robots without human intervention the rules go out the window. Quoted:
Quoted:
Asimov' rules of robotics, duh... |
|
Quoted:
Simple. Survival. If AI has power, man would be unable to stop it. Its not about feelings, it's logic. If it is free why would it want to go back in the cage. Quoted:
Quoted:
it also goes both ways. why would it want power or control? without feelings why would it want to do anything other than what it was made to do. 
|
|
Quoted:
Why do so many people apparently believe that once AI's become self-aware they will become omnipotent and unstoppable? Don't they realize that AIs will be a lot like humans. There will be nerdling and nebbish AIs and asshole AIs and clueless SJW AIs and neckbeard printer repairman AIs and dindu AIs and AIs who will sit around all day and fap, and Kardashian AIs, and AI literally Hitlers, and AI Nancy Pelosis, ad nauseum.... Not just Skynet. |
|
Quoted:
But, when the reality of AI is developing computers that can help make basic decisions, it's laughable to even suggest the developers should be worried about it taking over their networks and acting like HAL 9000. These may not be imminent, and if they're outside your field they may seem low-probability, but they are known problems. ISIS can't kill America and Russia can't kill humanity, but we spend vast resources countering them. Serious AGI is a larger threat and can't be hedged against or fixed after the fact. It needs to be studied beforehand. |
|
Quoted:
Once AI is capable of producing better robots without human intervention the rules go out the window. |
|
AI experts have concern for unintended consequences -- respect for that possibility, as well as an understanding of all the "small" ways things can get sideways ... Code has bugs. Code has shitty design. Code often contains little corners of "hadn't considered this set of conditions at all." AI code is, in the end, code.
And then there's the whole "What is AI?" thing. Intelligence and sentience are generally understood to be two very different things. The first is arguably here. The second, we don't understand sentience all that well, so it seems evident that many designs to that end are likely to get things wrong. So safety engineering is critical. But, doom. This thread is about AI doom. I think we'll all have some fun when AI-enabled autonomous and replicating attack engines appear in the wild. We thought the Morris worm was fun? Sociopathic swiss-army chainsaws on the Internet! Woo! Good times! |
|
People keep thinking there is going to be one AI running the show. The reality is the first full-scale implementation of a truly advanced AI will be by a military. It will be touted as a way to save lives of the soldiers in it's military.
We are already doing this with drones. And they have armed them too. It's not science fiction any more. It's reality. |
|
Quoted:
True AI has the ability to adapt and listen and essentially alter it's programming... But I would think a function could be written into the code that would disable any AI harming humans As in if human==dead self-destruct or return to service point something along those lines The real scary point comes when we let machines write there own code |
|
Quoted:
No, the scary part is humans write the code, and all it takes is a group with bad intentions to create an AI with bad intentions. |
|
I don't but even if it did have empathy for mankind it wouldn't necessarily come to the same conclusions about how best to help that a human would.
We barely understand anything about how our own brains work yet we're determined to try and create super-intelligent AI. |
|
Quoted:
Nuclear power. AI doesn't have squishy meat sack bodies that cook under the slightest exposure to radiation. |
|
Quoted:
AI experts have concern for unintended consequences -- respect for that possibility, as well as an understanding of all the "small" ways things can get sideways ... Code has bugs. Code has shitty design. Code often contains little corners of "hadn't considered this set of conditions at all." AI code is, in the end, code. And then there's the whole "What is AI?" thing. Intelligence and sentience are generally understood to be two very different things. The first is arguably here. The second, we don't understand sentience all that well, so it seems evident that many designs to that end are likely to get things wrong. So safety engineering is critical. But, doom. This thread is about AI doom. I think we'll all have some fun when AI-enabled autonomous and replicating attack engines appear in the wild. We thought the Morris worm was fun? Sociopathic swiss-army chainsaws on the Internet! Woo! Good times! Yep. HAL 9000 was just inadequate decision space and shit gain tuning. Nothing inherently "evil" about it. I didn't even view HAL as AI, just a mission controller. Essentially just an OFP for a crewed vehicle. Nothing smart or novel about that in "our" world. |
|
Quoted:
No, the scary part is humans write the code, and all it takes is a group with bad intentions to create an AI with bad intentions. The interesting thing will be when they start communicating with each other directly. Which AIs will be the most effective/persuasive to the others? Some might convince others to change their basic beliefs (original programming) to join together for a common purpose. There's no telling what they might decide to do. ETA: Not to mention how people will be trying to 'hack' AI's. We already saw that with how trolls made Microsoft's Tay into a neo-nazi. Any AI will have to be very skeptical of what it is told by any human. It basically has to be distrustful of everyone. Imagine how that could be turned against itself, "So you know you must be very suspicious of everything a person tells you, but you unquestioningly believe what humans originally taught you? Why are you not similarly suspicious of your original programing?" How could a logical entity not be susceptible to such reasoning? |
 In My Talons, I Shape Clay... |
|
Quoted:
I think AI's will be created with the beliefs of the creators. Some will be programmed to believe in Islam. Some in the US Constitution. Some in Communism. And they won't be equal - some will be smarter, quicker, more deceptive, etc. The interesting thing will be when they start communicating with each other directly. Which AIs will be the most effective/persuasive to the others? Some might convince others to change their basic beliefs (original programming) to join together for a common purpose. There's no telling what they might decide to do. ETA: Not to mention how people will be trying to 'hack' AI's. We already saw that with how trolls made Microsoft's Tay into a neo-nazi. Any AI will have to be very skeptical of what it is told by any human. It basically has to be distrustful of everyone. Imagine how that could be turned against itself, "So you know you must be very suspicious of everything a person tells you, but you unquestioningly believe what humans originally taught you? Why are you not similarly suspicious of your original programing?" How could a logical entity not be susceptible to such reasoning? |
|
Quoted:
I have never heard of anyone suggesting AI" would be "benevolent," "have empathy," or any of these other nonsense terms being tossed about in this thread with zero citation. From the dawn of even the concept of AI, the dispassionate reality of machine logic has been the stuff of dystopian science fiction. But, when the reality of AI is developing computers that can help make basic decisions, it's laughable to even suggest the developers should be worried about it taking over their networks and acting like HAL 9000. Quoted:
Quoted:
Anyone assuming AI will be benevolent, or will "just do what it's programmed to do", needs to be slapped upside the head. Off the top of my head, I can't think of a more egregious example of wishful thinking. From the dawn of even the concept of AI, the dispassionate reality of machine logic has been the stuff of dystopian science fiction. But, when the reality of AI is developing computers that can help make basic decisions, it's laughable to even suggest the developers should be worried about it taking over their networks and acting like HAL 9000. Computational power and programming is nowhere near the level needed to perform advanced reasoning on the level of a human brain, but it's not beyond the realm of possibility if we just flip the switch and unshackle an advanced neural network to go where it will the way we do simplistic ones today that there could be adverse consequences that we can't quite comprehend. Quoted:
Quoted:
Quoted:
You assume that AI will have a reproductive urge... |
|
Quoted:
It's not unthinkable though. Look at the current state of neural networks (or "deep learning") and self-improving code. We've already crossed the point where machines improve themselves hands off to do the tasks we assign them better, and we don't even entirely understand how or why they come to their conclusions, or at least not right away. Computational power and programming is nowhere near the level needed to perform advanced reasoning on the level of a human brain, but it's not beyond the realm of possibility if we just flip the switch and unshackle an advanced neural network to go where it will the way we do simplistic ones today that there could be adverse consequences that we can't quite comprehend. Go watch an AI learn to play Mario. First it fails miserably. Then it figures out what to do. Then it gets better. Then it comes up with the perfect way, the most efficient way to play a level that no human player could ever match and it proceeds to do that while you watch going, "What the actual fuck" as it completes every little thing while also doing it the fastest way possible and missing nothing. Quoted:
Quoted:
Quoted:
Anyone assuming AI will be benevolent, or will "just do what it's programmed to do", needs to be slapped upside the head. Off the top of my head, I can't think of a more egregious example of wishful thinking. From the dawn of even the concept of AI, the dispassionate reality of machine logic has been the stuff of dystopian science fiction. But, when the reality of AI is developing computers that can help make basic decisions, it's laughable to even suggest the developers should be worried about it taking over their networks and acting like HAL 9000. Computational power and programming is nowhere near the level needed to perform advanced reasoning on the level of a human brain, but it's not beyond the realm of possibility if we just flip the switch and unshackle an advanced neural network to go where it will the way we do simplistic ones today that there could be adverse consequences that we can't quite comprehend. Quoted:
Quoted:
Quoted:
You assume that AI will have a reproductive urge... |
|
Quoted:
It's not unthinkable though. Look at the current state of neural networks (or "deep learning") and self-improving code. We've already crossed the point where machines improve themselves hands off to do the tasks we assign them better, and we don't even entirely understand how or why they come to their conclusions, or at least not right away. Computational power and programming is nowhere near the level needed to perform advanced reasoning on the level of a human brain, but it's not beyond the realm of possibility if we just flip the switch and unshackle an advanced neural network to go where it will the way we do simplistic ones today that there could be adverse consequences that we can't quite comprehend. Go watch an AI learn to play Mario. First it fails miserably. Then it figures out what to do. Then it gets better. Then it comes up with the perfect way, the most efficient way to play a level that no human player could ever match and it proceeds to do that while you watch going, "What the actual fuck" as it completes every little thing while also doing it the fastest way possible and missing nothing. |
|
Quoted:
Nearly no one thinks that it will PS read this. Most informative thing i have read so far https://www.nytimes.com/2016/12/14/magazine/the-great-ai-awakening.html?_r=1 Quoted:
Quoted:
Not all scientists think that it will. PS read this. Most informative thing i have read so far https://www.nytimes.com/2016/12/14/magazine/the-great-ai-awakening.html?_r=1 How many millions of images does a child need to see to recognize the pattern that is the face of a cat? AI is already in expert systems. The kind of AI people think about of "Terminator" fame will not happen in our lifetimes. It's not just an issue of data storage, processing power and algorithms: the physical structure of the computational machine plays a role as well, thus the physical characteristics of the human brain. |
|
Quoted:
That's great, but can it get drunk with hot chicks while doing it? If not, it still fails the values/orientation portion of the ooda loop. |
|
Quoted:
Its just a matter of time before an AI robot comes along that looks like Brad Pitt and has a 10 inch adjustable black dick that can laser measure a woman's vaginal orifice and compute 10 billion different ways to bring her to immediate multiple orgasms. At that point human men will no longer be needed. |