It's sometimes argued that artificial intelligences cannot be accorded rights, or respect, because they cannot suffer. The argument goes that the suffering of an AI is unreal because it could be reprogrammed or deleted, thus expunging the experience.
I find this view horrific and I hope that I won't be alone in protesting bitterly if something like Westworld (the theme park, or the hosts in it) were ever to be made. What are we to determine what is and isn't deserving of our compassion?
I think that if I perceive the other entities to be capable of love, hope, creating relationships, kindness and joy I would object, virtual or not. The physicality of the entities may well lead me to have different beliefs about them, I think it would be hard not to be influenced by that in the way that a child is influenced by the big eyes and fur of a cuddly toy.
It's important to make this distinction between how you feel and what is actually occurring.
Supposing I made a robot baby that behaved exactly like a real one but with no AI would you feel differently? Would you feel differently if you knew.. versus didn't know?
My opinion on shows like Westworld is they are a form of mental-masturbation, designed to question our own morality. Unfortunately sometimes people get confused and project their own mortality on inanimate objects like robots.
Compassion is good, but projecting human ideas of suffering onto robots is the sort of thinking that gets you knitting sweaters for poor freezing penguins.
If you can explain the mechanism behind the manifestation then you can understand why it isn't real.
To me it's the understanding of why it isn't real that makes it not real. If you don't understand why the things you are brutalizing aren't real then you might as well be brutalizing small animals and I think it should be verboten until you are able to understand the consequences (or lack thereof) of your actions.
We are happy to brutalize animals for food or exterminate them as pests, but it's illegal to do it for pleasure (with many details and caveats). Going back to my long dormant religious education this is because how you treat powerless animals reflects the standards for how you will treat powerless humans. Society has a vested interest in keeping boundaries intact.
I would not hesitate to engage in all kinds of experimental and trolling behavior in Westworld confident in the knowledge that this is not how I would treat living creatures. I'm a gamer so exploring virtual worlds and systems in unconventional ways is just part of the fun. Putting those systems into meatspace sans meat doesn't change that.
So if we could explain consciousness would that mean brutalizing humans would be ok? I ask this as person with no religious beliefs (agnostic). I don't get at all how the organic matter in my skull results in what I (and I assume others) perceive as consciousness.
I would say no, but not as a question of morals or ethics just as a matter of practicality. I don't want to pick a fight with the world. Even if I could explain why I feel what I feel I am still programmed for self interest and to avoid pain and I see no reason to fight that programming.
> If you can explain the mechanism behind the manifestation then you can understand why it isn't real.
I watched, with my family, the Algorithms documentary available on Netfix. While not perfect (but very, very good), it certainly did a great job of explaining in simple terms some algorithms and how they relate to the world.
One of the most interesting parts, in my opinion, was the admittance during the section where they were discussing the XBox Kinect. It was all done through some machine learning algorithms, and impressively, they confessed to not being completely certain how the code actually worked. You could argue that this means that they cant explain the mechanism behind the manifestation.
Now, of course, this is a massively simplified version against the near-future Westworld narrative. However, it does call into question whether we would understand the mechanism of something so advanced. Also, applied to humans, if we end up understanding the mechanisms behind the brain, then does that make us less real?
They point out in episode three I think that one of the 'Hosts' are experiencing physical discomfort while going through diagnosis. If such a thing were coded, and it was made possible to respond to it (I would argue that the best possibility for making this possible would be a learning algo, similar to the Kinect analogy), then how would that be different from the very physical and chemical responses that a human has to experiencing suffering?
It's hard for me to make a decision in either direction. It is a complicated and nuanced problem. It reminds me of the short story "They're Made Out of Meat" [1], and could just be the result of human-centric (or in this case, biologically-centric) thinking.
There's an axiom of software engineering that any sufficiently large system (or, I suppose, algorithm) is incomprehensible by a single person. That's to say, a single person may be a high-level architect and understand how the high-level modules work together, but he won't understand how the inner workings of those modules work, especially if there are multiple layers to those modules, both bureaucratically and technically. And somebody who busies themselves understanding and working on the lowest levels, won't have the time to work on the higher architecture, nor would it interest him, as long as boundaries are well established and respected.
It's not that far of a stretch at all to claim that strong AI would be similar. At some point, everything worked on by today's AI researchers becomes some refined lower-level module, and tomorrow's researchers will use them as building blocks for the architecture for strong AI. Those lower-level modules will have coefficients derived from massive training sets, and tomorrow's researchers will simply understand that those coefficients work, without any real understanding of how those coefficients were derived. And that's OK.
I would probably be repelled or shocked in the moment, but I can imagine myself putting aside their appearance. I admit that I would probably continue to be disquieted by a horrifying appearance and would struggle to remember that such an entity had a life of the mind.
Do you mean "knowledge about yourself", such as realizing that you are a person, knowing your limitations, getting the word "I" right, predicting your own behavior, etc.? If so, that seems pretty easy to test. If someone/something can accurately answer detailed questions about themselves, obviously they are knowledgeable about themselves.
Do you mean the little "theater" we live in, where it feels like we're surrounded by sights and sounds and we're a little person in our head that chooses what to pay attention to? There's a book about this, and a proposed part of the brain that gives rise to it, though it's still fairly preliminary research. [1]
Do you mean our "inner narative" where we monologue to ourselves, slowly building up our "narrative identity"? [2]
Do you mean something else? These are all very different phenomenon, and you can have some without others, so I don't think a discussion can be very helpful until we've decided which we're talking about.
I really don't know the answer. When I use "self-aware" or "sentience" in regards to AI, I mean something along the lines of "does the AI posses "knowledge about itself" and does it have some form of inner monologue.
I think it's probably a blurry line. Great apes are probably just on the non-sentient, non-self-aware side, but boy are they close by some measures.
I'm trying to argue that these are largely unrelated questions. You can ask weather Great apes are self-aware in each sense, and get different answers:
(i) Great apes pass the mirror test, which certainly requires some level of self-knowledge.
(ii) Bernard Baars (a neuroscientist who studies consciousness) has a hypothesis about the part of the brain that gives rise to the "theater" of consciousness. I forget what part of the brain it is, but Great apes have it too, so if his hypothesis is correct then they also have a "theater".
(iii) Great apes don't speak, so they don't have anything like our inner monologue.
Honestly, I have no idea which of these three attributes an intelligent AI might have. And I see no reason to assume that they go together.
It's not a very nice thing, but I do not believe that any of the entities portrayed suffer for a moment. Nor do I believe that they have a life of the mind of any sort.
I think this is one of the many layers that make this show brilliant. Along with questions about sentience which while asked about the artificially created hosts, often pertain to human beings.
A bit Off-topic:
Definitively watch Westworld if you have some time to spare. Only three episodes aired so far and every single one of them was amazing and worth to watch IMHO.
Honestly I find the narrative rather boring. And it's not well-styled visually as for example Fargo or HBO's own The Night Of. Other than CGI I find there not much else that would captivate me or make me wanna dig into it's mysteries (admittedly I'm not a sci-fi fan).
This is the sort of sci-fi show that I feel I've been waiting for my entire life. I hope it doesn't go off the rails but so far it's been very deeply haunting and thought provoking.
30 comments
[ 3.2 ms ] story [ 56.0 ms ] threadI find this view horrific and I hope that I won't be alone in protesting bitterly if something like Westworld (the theme park, or the hosts in it) were ever to be made. What are we to determine what is and isn't deserving of our compassion?
How authentic must the experience be (or how sophisticated the AI) before you would object?
We are ourselves? Seems like the perfect individual to decide who deserves our compassion.
Supposing I made a robot baby that behaved exactly like a real one but with no AI would you feel differently? Would you feel differently if you knew.. versus didn't know?
My opinion on shows like Westworld is they are a form of mental-masturbation, designed to question our own morality. Unfortunately sometimes people get confused and project their own mortality on inanimate objects like robots.
To me it's the understanding of why it isn't real that makes it not real. If you don't understand why the things you are brutalizing aren't real then you might as well be brutalizing small animals and I think it should be verboten until you are able to understand the consequences (or lack thereof) of your actions.
We are happy to brutalize animals for food or exterminate them as pests, but it's illegal to do it for pleasure (with many details and caveats). Going back to my long dormant religious education this is because how you treat powerless animals reflects the standards for how you will treat powerless humans. Society has a vested interest in keeping boundaries intact.
I would not hesitate to engage in all kinds of experimental and trolling behavior in Westworld confident in the knowledge that this is not how I would treat living creatures. I'm a gamer so exploring virtual worlds and systems in unconventional ways is just part of the fun. Putting those systems into meatspace sans meat doesn't change that.
So if we could explain consciousness would that mean brutalizing humans would be ok? I ask this as person with no religious beliefs (agnostic). I don't get at all how the organic matter in my skull results in what I (and I assume others) perceive as consciousness.
I would say no, but not as a question of morals or ethics just as a matter of practicality. I don't want to pick a fight with the world. Even if I could explain why I feel what I feel I am still programmed for self interest and to avoid pain and I see no reason to fight that programming.
I watched, with my family, the Algorithms documentary available on Netfix. While not perfect (but very, very good), it certainly did a great job of explaining in simple terms some algorithms and how they relate to the world.
One of the most interesting parts, in my opinion, was the admittance during the section where they were discussing the XBox Kinect. It was all done through some machine learning algorithms, and impressively, they confessed to not being completely certain how the code actually worked. You could argue that this means that they cant explain the mechanism behind the manifestation.
Now, of course, this is a massively simplified version against the near-future Westworld narrative. However, it does call into question whether we would understand the mechanism of something so advanced. Also, applied to humans, if we end up understanding the mechanisms behind the brain, then does that make us less real?
They point out in episode three I think that one of the 'Hosts' are experiencing physical discomfort while going through diagnosis. If such a thing were coded, and it was made possible to respond to it (I would argue that the best possibility for making this possible would be a learning algo, similar to the Kinect analogy), then how would that be different from the very physical and chemical responses that a human has to experiencing suffering?
It's hard for me to make a decision in either direction. It is a complicated and nuanced problem. It reminds me of the short story "They're Made Out of Meat" [1], and could just be the result of human-centric (or in this case, biologically-centric) thinking.
1: http://www.terrybisson.com/page6/page6.html
It's not that far of a stretch at all to claim that strong AI would be similar. At some point, everything worked on by today's AI researchers becomes some refined lower-level module, and tomorrow's researchers will use them as building blocks for the architecture for strong AI. Those lower-level modules will have coefficients derived from massive training sets, and tomorrow's researchers will simply understand that those coefficients work, without any real understanding of how those coefficients were derived. And that's OK.
Can a human be re-programmed? Not easily (at this time), but certainly we can lobotomize, drug, and maim to the point that memories are lost.
Would a sentient AI feel pain, or other emotions? Is that part and parcel of sentience? Possibly?
Do you mean "knowledge about yourself", such as realizing that you are a person, knowing your limitations, getting the word "I" right, predicting your own behavior, etc.? If so, that seems pretty easy to test. If someone/something can accurately answer detailed questions about themselves, obviously they are knowledgeable about themselves.
Do you mean the little "theater" we live in, where it feels like we're surrounded by sights and sounds and we're a little person in our head that chooses what to pay attention to? There's a book about this, and a proposed part of the brain that gives rise to it, though it's still fairly preliminary research. [1]
Do you mean our "inner narative" where we monologue to ourselves, slowly building up our "narrative identity"? [2]
Do you mean something else? These are all very different phenomenon, and you can have some without others, so I don't think a discussion can be very helpful until we've decided which we're talking about.
[1] https://en.wikipedia.org/wiki/Global_Workspace_Theory [2] https://en.wikipedia.org/wiki/Narrative_identity
I think it's probably a blurry line. Great apes are probably just on the non-sentient, non-self-aware side, but boy are they close by some measures.
(i) Great apes pass the mirror test, which certainly requires some level of self-knowledge.
(ii) Bernard Baars (a neuroscientist who studies consciousness) has a hypothesis about the part of the brain that gives rise to the "theater" of consciousness. I forget what part of the brain it is, but Great apes have it too, so if his hypothesis is correct then they also have a "theater".
(iii) Great apes don't speak, so they don't have anything like our inner monologue.
Honestly, I have no idea which of these three attributes an intelligent AI might have. And I see no reason to assume that they go together.