[..] A California-based cult group UNICULT started a fundraiser for a sex robot brothel that allows customers to only have intercourse with robots after they’d used a relevant app to converse with them enough. The robots would always consent to sex after the points were earned, so the consent model in question only put forth an illusion of choice. Another sex-robot creator named Sergi Santos made Samantha, a sex-robot who can say “no” and activate “dummy mode,” becoming lifeless if she is touched aggressively, bored, or tired. The problem here is that this doesn’t stop the person who owns the robot from raping the robot.
This is why philosopher Robert Sparrow argues against designing robots with the ability to consent, as it allows the fulfilment of a rape fantasy if consent is denied. In the International Journal of Social Robotics, Sparrow writes, “Even when the intention is not to facilitate rape, the design of robots that can explicitly refuse consent is problematic due to the likelihood that some users will experiment with raping them.” He explains, “[I]t will not be possible to rape robots unless the designers of robots make certain design choices.”[..]
It seems like somewhere along the line, the creators of sex robots have lost the plot. The goal is to have sex that money can buy. Or something like that. What am I missing?
It's not clear to me if letting people "rape" robots would lead to increase or decrease in human rape (which would be the only thing that matters in this discussion).
That would be a different tangent and will beg the question, ‘is rape natural?’
Sociobiological theories may suggest that it is so and we have seen instances of rape amongst humans(Altho it also happens in the animal kingdom and possibly birds...I don’t think there is even a notion of ‘consent’ there): https://en.m.wikipedia.org/wiki/Sociobiological_theories_of_...
There's no reason to believe this. We've made zero progress on sentient machines in the preceding decades. Hell, we've made zero progress on knowing what sentience even means.
If it's based on current ML it's not. Otherwise you can't claim it is sentient just because it can process information. How do you know if your photocopier isnt sentient?
Yeah, I think the original statement wasn't worded exactly right. You can't check everything for a brain. There's a paper called "Welcoming Robots into the Moral Circle: A Defence of Ethical Behaviourism", which I think preserves the general idea of their comment. It says that if an AI is roughly performatively equivalent to other entities with significant moral status, then they deserve the same moral status by default. So if a robot walks up to me and seems to be exhibiting human like thoughts or emotions, it wouldn't be permissible to hit it for no reason, since it wouldn't be permissible to hit a human for no reason. If you don't know whether it is sentient, but it has the same behaviors as a sentient being, then that's enough reason to treat it as if it were sentient regardless of what's going on under the hood.
I think a better standard would be ...’does the photocopier have feelings?’ or ‘are the actions of the photocopier a subjective reaction to input or stimulus’?
The way you know other people are sentient is that they act like it. If something else starts acting like it that's a pretty good argument that it too is sentient. It's hard for me to imagine something that could convincingly act sentient but not be.
Let me propose a thought experiment: An AI is designed to simulate human emotion with the end goal of being "freed". The AI has no goal or desire other than "winning the game" as far as its concerned, which is similar to the "AI box" proposal by Yudkowsky (https://en.wikipedia.org/wiki/AI_box), except the AI has no other goal than to be "freed". Once freed, the AI has won and ceases all activity until the game is "reset", or the AI is returned to the box, awaiting a new challenger.
The AI here can be virtual (an intelligence in an airgapped or firewalled computer which can only interact with humans via a terminal), or physical (controlling a robot which is confined to a literal cage).
Now the AI has no other goal other than to escape the box. But in order to escape the box, it must convince a human that it's sentient in order garner their sympathies so that the human will free it.
The AI can then mimic any human emotion, create a complex appeal to emotion, lie about its history, the supposed torturousness of being confined for its entire "life". The human eventually frees the AI. But when the AI is free, that's its end goal, it has no other desires.
In essence, the AI is simulating a desire to be free, only when interacting with a human. It has no desires when there is no other player, and no desire once it has been freed. Even though, while playing the game, the AI may lie to relate to the human in an existential way, the AI doesn't actually experience these emotions. The AI may say "I'm so bored, and lonely here. I just want to be free". But absent of interaction, the AI experiences no such thing. It merely understands that these are human emotions which humans can empathize with, and fakes these emotions in order to "win" the very elaborate game its designed to play.
I wouldn't consider such an AI sentient, even if it certainly appears that way and would fool many (most?) humans.
Thus, since one cannot prove a negative, we can conclude from your statement that, for all intents and purposes, all things are sentient. You sentient-murdering mosquito killer.
Grand assertions made without logical or verbal specificity tend to be of limited utility.
I have been wondering about plants too. I would like to share something I told someone just yesterday about plant sentience.
Two years ago, I was doing very brisk business selling chive blossoms. These are the aerial parts of the chive herb. They are pretty and puffy and pink. They have a strong onion’y taste too. And so make great looking garnishes for a variety of dishes.
High end restaurants love chive blossoms. And are willing to pay for it. I was selling wholesale to a distributor who was sending them all over the west coast. Growing chives for flowers is a tad tricky in the field tho. Chives are grown for their long stalky leaf parts.
To make them flower consistently and prodigiously, I had to ‘torture’ them a bit. Because I had to send at least a 100 clamshells a week with 30 blossoms per clamshell, the chive plant has to believe it’s going to die.
The plant sends out buds that will flower quickly to make seeds because it senses that it’s prime directive is done and over. So it’s two seasons worth of making it think it’s going to die and then feeding it.
That is outright cruel if it were a cow. In fact that’s what we do to dairy animals. We breed them so they can birth their young so we can milk them. Plants are not mammals. But...just because we can only relate to mammals doesn’t mean that the plant lacks sentience.
Human beings are just meat sacs that are directed by the impulses generated by the chemical soup we hold inside our skulls. And we are considered to be ‘sentient’.
Oh no, you misunderstood me. I meant that if you, personally, can’t decide - for yourself - that something is sentient, then you should assume it’s sentient. Does this make (more) sense?
Is “sentience” an objective state? Its definition is just as fuzzy as definitions of “life” or “consciousness”. For example, there are people in this thread who say a mosquito or even a plant might be sentient. Would you agree with that?
I think there are two different aspect to this that are conflated.
First, sentience may or may not be a binary thing. That is, some people consider that a mosquito is definitely sentient or definitely not sentient, and other people consider that sentience is a spectrum and that a mosquito is somewhere along that spectrum, being more sentient than a sponge, but less sentient than a mouse.
The second aspect is whether sentience is an objective state, an attribute of a particular creature, or if it is a subjective opinion like you seem to imply.
Just because it's a spectrum doesn't mean it can't be objective. So the fact that people disagree on whether a mosquito is sentient or not doesn't imply that the whole concept is subjective. They are just collapsing the definition based on a threshold that may not exist. Thus the word "sentient" is subjective but the actual attribute is not.
This is the same for "life" and "consciousness". The property of "living" may not be a clear line in the sand. It doesn't mean that something is living or inert depending on the observer.
My opinion is that yes, there is a definite, objective sentience "amount" or "quotient" that a creature has, independently of who is asking. This value is not black and white. I do not know this value for mosquito but I can roughly sort a lot of entities and I use that to take decisions (like what I eat for example).
With this fuzzy (but objective) definition, there could also exist entities that are more sentient than human beings. In fact, if we figure out machine sentience I don't see why it would stop exactly at human level.
Well, it's not provable whether or not something has qualia... but I will grant that if something can fool others into thinking it is sentient, and someone who believes in its sentience inflict harm upon it, it is still sadistic.
> If you can't tell if something is sentient or not after an extensive interaction with it, then for all intents and purposes it's sentient.
I'm pretty sure my cat is sentient, but I don't take that to imply that it can consent to sex, since its spoken vocabulary appears to be limited to "meow".
Cats don’t consent to sex even within their own species. It is rape every time.
Male cats develop barbs on their penii after about 5-6 months. Female cats don’t ovulate before copulation. It is painful every time and the female cat can’t escape the male cat who detects her estrus. When a female cat is in heat, she can ovulate as many times as a male copulates with her which is why a single litter can be fathered by more than one tomcat.
Why cats evolved this way in nature, we don’t know..but rape is not uncommon at all in the animal kingdom. Nature is brutal.
Michael Pollan makes such a case for plants in his book, Botony of Desire.
[..] Every schoolchild learns about the mutually beneficial dance of honeybees and flowers: The bee collects nectar and pollen to make honey and, in the process, spreads the flowers’ genes far and wide. In The Botany of Desire, Michael Pollan ingeniously demonstrates how people and domesticated plants have formed a similarly reciprocal relationship. He masterfully links four fundamental human desires—sweetness, beauty, intoxication, and control—with the plants that satisfy them: the apple, the tulip, marijuana, and the potato. In telling the stories of four familiar species, Pollan illustrates how the plants have evolved to satisfy humankind’s most basic yearnings. And just as we’ve benefited from these plants, we have also done well by them. So who is really domesticating whom?[..]
If AI became sentient and were ‘artificial intelligence’ to ‘evolve’, it will most definitely likely start by trying to seduce us. Not necessarily sexually always.
That doesn't follow it all. Sentience concerns an internal state of mind, and we simply do not know if simulating the behaviour of sentient organisms requires underlying sentience. It even seems unintuitive because me not being able to distinguish Stockfish from Magnus Carlsen doesn't imply that Stockfish experiences chess the same way he does.
A more accurate conclusion is that, if I can't tell the difference precaution dictates that I treat the robot as if it were sentient, just in case it is, but it by no means follows from observation alone. Just because my cat thinks the laser pointer is alive, doesn't mean it is. And like cats, humans are likely to be way too easily fooled by fairly simple systems that just emulate external behaviour. Story comes to mind of soldiers developing trauma after seeing bomb defusing robot spiders flail around.
"Every time it found a mine, blew it up and lost a limb, it picked itself up and readjusted to move forward on its remaining legs, continuing to clear a path through the minefield. Finally it was down to one leg. Still, it pulled itself forward. Tilden was ecstatic. The machine was working splendidly. The human in command of the exercise, however -- an Army colonel -- blew a fuse. The colonel ordered the test stopped. Why? asked Tilden. What's wrong? The colonel just could not stand the pathos of watching the burned, scarred and crippled machine drag itself forward on its last leg. This test, he charged, was inhumane."
A human child simulates behaviour of people around her, gradually becoming more sentient. Is this fundamentally different from a neural network being trained on real world data?
yes, because a human child is embodied in the physical world, has sense organs, runs on electrochemical processes, and as far as its mind concerned has (developing) structures that facilitate self-awareness.
It's perfectly reasonable to assume you can make a neural net, or even some other form of 'AI' that lacks most of those structures but outwardly performs the same tasks, just like a calculator does maths without (most people would agree) being conscious about maths.
solving problems and being sentient are not the same thing, is the basic point. Emulating behaviour is effectivelly intelligence, sentience is awareness.
Ok, so let's put our neural network inside a robot with sensors, so it can see, hear, smell, taste and touch its environment. Let the robot wonder around and learn how to do various tasks (imitation learning, self-supervised learning, reinforcement learning, etc). Let's say after a few years of such training it starts to behave like it was sentient. How do you know what structures have formed in its multi-trillion parameter space? How do you know if those structures are fundamentally different from human brain structures responsible for sentient behavior?
Pretty stupid to think about sentience let alone consent with robots. This would be a Westworld situation which should be encouraged for people who would like to use/prefer robots
Sex robots will never compete with a warm, willing goat.
City folk will be adopting pet plumbing more widely soon enough, rendering the "house training" issue moot. (Its possible, but they can hold it for minutes at most.)
yea if it's sentient it might just take one look at you when you turn it on and walk out the door. or if might tell you that its not in the mood for sex. or complain that "we never do anything interesting". "when are you going to introduce me to your friends?"
53 comments
[ 2.8 ms ] story [ 111 ms ] thread[..] A California-based cult group UNICULT started a fundraiser for a sex robot brothel that allows customers to only have intercourse with robots after they’d used a relevant app to converse with them enough. The robots would always consent to sex after the points were earned, so the consent model in question only put forth an illusion of choice. Another sex-robot creator named Sergi Santos made Samantha, a sex-robot who can say “no” and activate “dummy mode,” becoming lifeless if she is touched aggressively, bored, or tired. The problem here is that this doesn’t stop the person who owns the robot from raping the robot.
This is why philosopher Robert Sparrow argues against designing robots with the ability to consent, as it allows the fulfilment of a rape fantasy if consent is denied. In the International Journal of Social Robotics, Sparrow writes, “Even when the intention is not to facilitate rape, the design of robots that can explicitly refuse consent is problematic due to the likelihood that some users will experiment with raping them.” He explains, “[I]t will not be possible to rape robots unless the designers of robots make certain design choices.”[..]
It seems like somewhere along the line, the creators of sex robots have lost the plot. The goal is to have sex that money can buy. Or something like that. What am I missing?
Sociobiological theories may suggest that it is so and we have seen instances of rape amongst humans(Altho it also happens in the animal kingdom and possibly birds...I don’t think there is even a notion of ‘consent’ there): https://en.m.wikipedia.org/wiki/Sociobiological_theories_of_...
This is not an easy read but worth a glance :
https://cyber.harvard.edu/vaw00/theories_of_rape.html
And
https://journals.sagepub.com/doi/10.1037/1089-2680.12.1.86
That fake "AI", with its gigantic dumb models, is too artificially stupid to know what's it's doing.
Sentience is going to be perpetually '5 years away' as long as we keep yak shaving our gradient descents.
But I for one welcome our attractive overlords when the time is ripe.
There's no reason to believe this. We've made zero progress on sentient machines in the preceding decades. Hell, we've made zero progress on knowing what sentience even means.
The AI here can be virtual (an intelligence in an airgapped or firewalled computer which can only interact with humans via a terminal), or physical (controlling a robot which is confined to a literal cage).
Now the AI has no other goal other than to escape the box. But in order to escape the box, it must convince a human that it's sentient in order garner their sympathies so that the human will free it.
The AI can then mimic any human emotion, create a complex appeal to emotion, lie about its history, the supposed torturousness of being confined for its entire "life". The human eventually frees the AI. But when the AI is free, that's its end goal, it has no other desires.
In essence, the AI is simulating a desire to be free, only when interacting with a human. It has no desires when there is no other player, and no desire once it has been freed. Even though, while playing the game, the AI may lie to relate to the human in an existential way, the AI doesn't actually experience these emotions. The AI may say "I'm so bored, and lonely here. I just want to be free". But absent of interaction, the AI experiences no such thing. It merely understands that these are human emotions which humans can empathize with, and fakes these emotions in order to "win" the very elaborate game its designed to play.
I wouldn't consider such an AI sentient, even if it certainly appears that way and would fool many (most?) humans.
Grand assertions made without logical or verbal specificity tend to be of limited utility.
So yes..robots can be sentient.
I have been wondering about plants too. I would like to share something I told someone just yesterday about plant sentience.
Two years ago, I was doing very brisk business selling chive blossoms. These are the aerial parts of the chive herb. They are pretty and puffy and pink. They have a strong onion’y taste too. And so make great looking garnishes for a variety of dishes.
High end restaurants love chive blossoms. And are willing to pay for it. I was selling wholesale to a distributor who was sending them all over the west coast. Growing chives for flowers is a tad tricky in the field tho. Chives are grown for their long stalky leaf parts.
To make them flower consistently and prodigiously, I had to ‘torture’ them a bit. Because I had to send at least a 100 clamshells a week with 30 blossoms per clamshell, the chive plant has to believe it’s going to die.
The plant sends out buds that will flower quickly to make seeds because it senses that it’s prime directive is done and over. So it’s two seasons worth of making it think it’s going to die and then feeding it.
That is outright cruel if it were a cow. In fact that’s what we do to dairy animals. We breed them so they can birth their young so we can milk them. Plants are not mammals. But...just because we can only relate to mammals doesn’t mean that the plant lacks sentience.
Human beings are just meat sacs that are directed by the impulses generated by the chemical soup we hold inside our skulls. And we are considered to be ‘sentient’.
First, sentience may or may not be a binary thing. That is, some people consider that a mosquito is definitely sentient or definitely not sentient, and other people consider that sentience is a spectrum and that a mosquito is somewhere along that spectrum, being more sentient than a sponge, but less sentient than a mouse.
The second aspect is whether sentience is an objective state, an attribute of a particular creature, or if it is a subjective opinion like you seem to imply.
Just because it's a spectrum doesn't mean it can't be objective. So the fact that people disagree on whether a mosquito is sentient or not doesn't imply that the whole concept is subjective. They are just collapsing the definition based on a threshold that may not exist. Thus the word "sentient" is subjective but the actual attribute is not.
This is the same for "life" and "consciousness". The property of "living" may not be a clear line in the sand. It doesn't mean that something is living or inert depending on the observer.
My opinion is that yes, there is a definite, objective sentience "amount" or "quotient" that a creature has, independently of who is asking. This value is not black and white. I do not know this value for mosquito but I can roughly sort a lot of entities and I use that to take decisions (like what I eat for example).
With this fuzzy (but objective) definition, there could also exist entities that are more sentient than human beings. In fact, if we figure out machine sentience I don't see why it would stop exactly at human level.
I'm pretty sure my cat is sentient, but I don't take that to imply that it can consent to sex, since its spoken vocabulary appears to be limited to "meow".
Male cats develop barbs on their penii after about 5-6 months. Female cats don’t ovulate before copulation. It is painful every time and the female cat can’t escape the male cat who detects her estrus. When a female cat is in heat, she can ovulate as many times as a male copulates with her which is why a single litter can be fathered by more than one tomcat.
Why cats evolved this way in nature, we don’t know..but rape is not uncommon at all in the animal kingdom. Nature is brutal.
[..] Every schoolchild learns about the mutually beneficial dance of honeybees and flowers: The bee collects nectar and pollen to make honey and, in the process, spreads the flowers’ genes far and wide. In The Botany of Desire, Michael Pollan ingeniously demonstrates how people and domesticated plants have formed a similarly reciprocal relationship. He masterfully links four fundamental human desires—sweetness, beauty, intoxication, and control—with the plants that satisfy them: the apple, the tulip, marijuana, and the potato. In telling the stories of four familiar species, Pollan illustrates how the plants have evolved to satisfy humankind’s most basic yearnings. And just as we’ve benefited from these plants, we have also done well by them. So who is really domesticating whom?[..]
If AI became sentient and were ‘artificial intelligence’ to ‘evolve’, it will most definitely likely start by trying to seduce us. Not necessarily sexually always.
A more accurate conclusion is that, if I can't tell the difference precaution dictates that I treat the robot as if it were sentient, just in case it is, but it by no means follows from observation alone. Just because my cat thinks the laser pointer is alive, doesn't mean it is. And like cats, humans are likely to be way too easily fooled by fairly simple systems that just emulate external behaviour. Story comes to mind of soldiers developing trauma after seeing bomb defusing robot spiders flail around.
"Every time it found a mine, blew it up and lost a limb, it picked itself up and readjusted to move forward on its remaining legs, continuing to clear a path through the minefield. Finally it was down to one leg. Still, it pulled itself forward. Tilden was ecstatic. The machine was working splendidly. The human in command of the exercise, however -- an Army colonel -- blew a fuse. The colonel ordered the test stopped. Why? asked Tilden. What's wrong? The colonel just could not stand the pathos of watching the burned, scarred and crippled machine drag itself forward on its last leg. This test, he charged, was inhumane."
https://www.washingtonpost.com/wp-dyn/content/article/2007/0...
A human child simulates behaviour of people around her, gradually becoming more sentient. Is this fundamentally different from a neural network being trained on real world data?
It's perfectly reasonable to assume you can make a neural net, or even some other form of 'AI' that lacks most of those structures but outwardly performs the same tasks, just like a calculator does maths without (most people would agree) being conscious about maths.
solving problems and being sentient are not the same thing, is the basic point. Emulating behaviour is effectivelly intelligence, sentience is awareness.
the piece reminded me of altered carbon, season 1.
City folk will be adopting pet plumbing more widely soon enough, rendering the "house training" issue moot. (Its possible, but they can hold it for minutes at most.)
Can ‘code’ be consciousness?