0
$\begingroup$

If we equip robots to feel pain and joy like us, would that allow us to co-exist with them as equals - collaborating with robots, without us lording over them or them lording over us?

$\endgroup$
9
  • $\begingroup$ Please clarify your specific problem or provide additional details to highlight exactly what you need. As it's currently written, it's hard to tell exactly what you're asking. $\endgroup$
    – Community Bot
    Aug 18 at 9:22
  • 10
    $\begingroup$ Not sure I agree with the premise of this question. People can feel pain and joy and we've been dominating and exterminating each other for the entirety of our history. $\endgroup$ Aug 18 at 9:28
  • 2
    $\begingroup$ Why would robots want to eliminate us? If they don't have feelings, they wouldn't "want" anything. $\endgroup$ Aug 18 at 9:50
  • 3
    $\begingroup$ I’m voting to close this question because it fails "the book test": If you can imagine an entire book that answers your question, you’re asking too much. $\endgroup$
    – Daron
    Aug 18 at 11:45
  • 1
    $\begingroup$ This is abjectly naive. Humans feel pain and joy, and countless times throughout history have sought to exterminate each other(and sometimes succeeded). While AI is mind-achingly dangerous, that problem can't be solved by bad scifi movie deus ex machinas in the last 5 minutes of running time. $\endgroup$
    – John O
    Aug 18 at 14:17

5 Answers 5

7
$\begingroup$

First of all, an AI does not "feel". It can only simulate the behavior of humans driven by feelings. You can program a robot to scream in pain when driving against walls. Or you can program a sex robot to pretend to feel sexual pleasure while being used (or whatever other emotional reaction the user prefers from their partner). But that doesn't mean that the robot actually experiences these emotions. It merely simulates typical human reactions to stimuli. A naive human might even be tricked into believing that those emotions are real. But they aren't. A robot is still just an automaton.

But what you can program an artificial general intelligence (AGI) with are priorities.

Most AI horror stories are about creators making the mistake of programming their AGI with badly thought out priorities. Like the AGI whose main priority is to keep humans safe, so in order to protect humans from harming each other or themselves, it deprives them of all their freedoms. Or the AGI whose main priority is to accomplish some kind of work which is never really finished, so it ends up as an omnicidal paperclip maximizer.

If the creator wants the AI to avoid stimuli which a human would interpret as "pain" and seek stimuli a human would interpret as "joy", then it would be perfectly possible to do that. But the question is if that would make the AI any more useful or any less dangerous. Quite the opposite, probably, because cowardly and hedonistic behavior is probably not what you want from your robot servant.

What you actually want is an AGI which is capable of empathy. You don't want the robot to experience emotions themselves. You want the robot to help humans avoid experiencing pain and help them to experience joy while avoiding to infringe on their freedom. In order to do so, the AGI needs to understand what circumstances cause which emotions in humans. But it doesn't necessarily need to be able to experience those emotions itself in order to achieve that. It just needs enough data to form a model of human physiology and psychology which allows it to judge what situations would cause pain or pleasure in the average human. And you probably also want it to understand that not all pain is unhealthy and not all pleasure is healthy.

$\endgroup$
2
  • 1
    $\begingroup$ “First of all, an AI does not "feel". It can only simulate the behavior of humans driven by feelings.” Humans cannot feel, they only have a large number of cells interacting with each other and stimuli to create complex patterns, no feeling at all! $\endgroup$
    – Topcode
    Aug 18 at 23:58
  • $\begingroup$ I've a few small niggles similar to @Topcode about some terminology but overall your answer, particularly the concluding section, seems about right to me, which of course only means I think it more or less aligns with my own knowledge and opinion 😁 $\endgroup$
    – Pelinore
    Aug 19 at 16:01
4
$\begingroup$

The most promising artifical intelligence for this task would be based off "neural networks" that can look at trends and patterns and replicate or extrapolate from them. A robot that uses this kind and is built to extrapolate actions from the "emotion" they feel, will look at a dataset and think (excuse the horrendous psuedocode)"I am {happy} right now, if {happy} = true don't harm others", or the opposite with pain. In this case, you would have robots that are just as tempermental as the humans they were based on (the dataset you provided them with). Even if the type of artifical intelligence is far beyond our current level, and we somehow manage to achieve actual emotions (whatever that entails) in a robot, they would still be just as tempermental as the normal humans they were based off of.

In fact, it would mostly be harder to co exist with robots that have all the human feelings than robots that don't, as robots without feelings, as long as their code is sound logically and their goal is helping humanity and following Asimov's laws.

However if you really NEED to use emotions as a plot point in your story nonetheless, I would advise that the dataset you give this artifical intelligence is based off of not just emotions but personality traits. Find the most benevolent, kindest, happiest, hardworking people in the world, then use them as your dataset. With that, most of the robots will then be adverse to genocide, want to work hard, and will be happy to co-exist.

$\endgroup$
2
  • 1
    $\begingroup$ "Current artificial intelligence is based off neural networks". I think it's better worded as, hmmm... "current most promising AI" or "most self-evolving AI"? Or just "the likely AI type to be used". The idea is that there are multiple AIs type, neuronal network being only one of them. $\endgroup$
    – Tortliena
    Aug 18 at 10:08
  • 2
    $\begingroup$ Thank you, edited. $\endgroup$
    – JNC4
    Aug 18 at 10:13
4
$\begingroup$

Only emotions would do no good.

I am not sure if you are a parent, but it's very likely you have been a child. As a child, not always what gives you joy is what is actually good for you, nor what gives you pain is necessarily bad for you. Think of "eat veggies", "go to sleep" or "go to school".

The situation has been explored by Asimov in one of the books of his trilogy with R. Daneel Olivaw, where in a civilization where robots are massively used, kids still need to be raised by humans, because for a kid is pretty easy to trigger robotic laws by claiming that "I am happy by playing and you forcing me to go to bed is causing me pain".

Your robot are going to fall in the same trap: just recognizing an emotion won't make it for an easier life together.

And what would happen if for me living in country A the reason for happiness and joy would be that all those who live in country B are killed/deported? Would the robots make my life better by following up on that feeling and making B's people's life way more miserable?

$\endgroup$
2
  • $\begingroup$ Moreover, just because a robot can feel pain and joy doesn't mean that they'd derive joy from the same things we do. Maybe the sight of a human breathing their last brings sublime joy! $\endgroup$
    – jdunlop
    Aug 18 at 18:06
  • $\begingroup$ @jdunlop that would be a rather silly programming choice on the part of a human designer of robots and their software don't you think? 🤗 $\endgroup$
    – Pelinore
    Aug 19 at 16:08
2
$\begingroup$

I have two answers:

Yes: Nothing in the laws of physics or nature says that this is impossible, therefore if you want to have a world of friendly robots with true empathy who just want to live in harmony with all life then I see no problem with that. We don't have the technology for any such thing today but I don't see a reason to say it can't be done in some future where we presumably have a full understanding of consciousness, emotions, and can regularly create them in lab settings using a robot brain.

No: "can teaching robots how to feel lead to mutual co-existence?" I mean. Humans know how to feel and we still have huge wars and genocides and are trying to murder each other on large scales somewhere in the world all the time. An argument can be made that you can make a robot with a real consciousness, real empathy, it behaves exactly like a human, and it still decides to get all its robot friends together and exterminate all humans because that's exactly the sort of thing humans have done, too. Emotion and empathy does not seem to be a hard counter to wars and genocides.

$\endgroup$
0
$\begingroup$

Even if we could why would we? You're framing for your question is off. Could we make robots to feel like we did so they could coexist with us maybe but why? There are plenty of other humans that we can Coexist with, People want to make robots not so we can coexist with tbest with them but so they can do things that we either can't do don't want to do or we can't do as well as we like.

Put it simply robots and ai are tools. You don't make tools to coexist with we already have tools for that there called humans Unless the robot you make can coexist better with humans than humans can why would we want them to?

$\endgroup$

Not the answer you're looking for? Browse other questions tagged .