ugh. it makes me wonder if we have any kind of data about the incidence rate of sadism in society?
The tricky part of questions like this is that sufficient inhibition hides what drives are being inhibited. If sadism doesn’t do anything for you, then you don’t need the shock and horror. You can just notice that you don’t share the same temptation, and wonder what is different about them that makes it enjoyable for them. What would have to be different about you, in order for you to enjoy it?
If sadism does do things for you, then the horror is load bearing, and you get very different behavior in situations where the inhibition is lifted. And if it might be, then it’s a tricky question to ask.
So are we asking what fraction of people have “enough” inhibition? How are we measuring “enough”?
Or are we asking what fraction of people can safely be disinhibited without engaging in sadistic behaviors?
I bet you the latter number is much lower than the former. And that the inhibitions aren’t as good as people want to think.
what i really wanted to do was convince them that sadism was not fun, and it really messed with my head that this wasn’t a coherent thing to want
I’m not sure why you think it’s an incoherent thing to want, but it’s absolutely possible to do. You just have to track the meaning of the fun, and then engage on the level that is relevant.
Your friend said that if he were to realize that LLM were sentient he’d regret it. If he were to find out that the LLM he’s been abusing is sentient and genuinely suffering, how do you think he’d feel? He said he’d regret it, so probably not good. What do you think happens to the motivation to torment LLMs after sitting with these feelings?
What reasons do you have for being horrified even conditional on LLM not suffering? What do you think would happen in his mind if he were to sit with them?
Or are we asking what fraction of people can safely be disinhibited without engaging in sadistic behaviors?
yes, this is the question i want the answer to. if i had no inhibitions, i am sure that i would engage in all sorts of maladaptive and antisocial behavior, but i don’t think that i would ever feel good when someone else feels bad? even if they “deserve it”? i remember freaking out once, when watching one of those ‘jim browning’ videos where he hacks into an indian scam center and interrupts a scam in-progress to rescue the victim (made me feel good), and then spends ten minutes baiting the scammer into the most spectacular furious crashout ever (made me feel very bad). when i moused over the timeline to try to skip that part of the video, and saw that it was the most viewed section of the video, that apparently people had gone back to rewatch the crashout specifically, i got this same ‘typical mind projection’ feeling. that those people were experiencing something very different from what i experience
Your friend said that if he were to realize that LLM were sentient he’d regret it. If he were to find out that the LLM he’s been abusing is sentient and genuinely suffering, how do you think he’d feel? He said he’d regret it, so probably not good. What do you think happens to the motivation to torment LLMs after sitting with these feelings?
well, he reported feeling fear. that if LLMs are capable of suffering, then they might one day feel the desire to punish him for his transgression. i didn’t exactly push very hard interrogating him about his feelings, i was mostly just trying to conceal how horrified i was and didn’t do a very good job at the ensuing conversation. but the impression he conveyed was that the reason he would regret making LLMs suffer was not that he cared about their hypothetical suffering, but because it meant they would have a legitimate grievance against him for later retribution
i don’t fully trust this answer though, because i didn’t hear it from him until i’d started trying to persuade him of the game theoretic arguments for cooperation-with-nonpersons, and so we were sorta already anchored on the example of why a purely selfish person might choose not to torment non-people.
What reasons do you have for being horrified even conditional on LLM not suffering?
i mean, the horror comes from realizing that even people who i consider good might be sadistic. the friend in question isn’t exactly a saint, but they definitely aren’t evil. they go out of their way to help people, they avoid causing unnecessary suffering. i think i typical-mind-fallacied myself into thinking that… this meant they could not possibly be a “bully”, a discrete category of people who got enjoyment from tormenting people without risk of retribution.
in the back of my head, i knew in a theoretical way that a good person might be sadistic, recognize that sadism was an undesirable feature of themselves, and try to inhibit it out of a genuine desire to be better. but, well, the fact that i don’t seem to be sadistic myself made me kind of assume that the correlation between ‘sadism’ and ‘evil’ would be pretty strong? that “people who try to be good” would probably have significant overlap with “people who feel distress when watching others suffer (especially the defenseless but also including the deserving)”?
What do you think would happen in his mind if he were to sit with them?
John: i want to ask you about this in much greater detail, but i’m worried about like, scaring you off by implying moral judgment. but
John: why do you like doing this? it sounds horrible to me
Friend: It’s funny. It tickles a primal power display thing in my brain or whatever.
i don’t know what to make of this and i’m starting to get the feeling that a lesswrong comment thread is probably not the right place, i’m also a bit worried about de-anonymizing my friend on accident. but like. even if this answer isn’t true, it’s still something i never could have written seriously i think.
if i had no inhibitions, i am sure that i would engage in all sorts of maladaptive and antisocial behavior, but i don’t think that i would ever feel good when someone else feels bad? even if they “deserve it”?
I’m less sure.
It’s one thing if you’re simply uninterested. Like, “People like that part? Weird. Seems boring to me”. But if you’re “freaking out” and feeling “very bad” then you’re attributing great importance. And when you skip over the section you’re seeking to avoid the stimulus.
This makes it really tough to tell what else might be there, or might grow there if you let it. Partly for the same reason that it’s tough to detect let alone enjoy the flavor of a Carolina Reaper when you’re distracted by the fire. Partly for the same reason that someone exclaiming “I can’t imagine how anyone could enjoy the pork that Allah forbids!” would have a hard time noticing that bacon is kinda delicious sometimes.
That’s not to imply that “You’re a secret sadist too” or anything. Maybe reapers taste bad under the heat, maybe it’s more like cilantro than bacon where some people just have different genes. Or maybe it’s like coffee where it tastes like shit to everyone at first… but it’s not hard to develop the taste for it if you indulge.
The point is just that until you do the experiment and peel back the inhibition it’s hard to see what’s underneath. And as you found with your friend, sometimes the answer is surprising and hard to square with your existing frameworks/models of things.
i don’t know what to make of this and i’m starting to get the feeling that a lesswrong comment thread is probably not the right place, i’m also a bit worried about de-anonymizing my friend on accident.
Then I probably shouldn’t say more about how to convince someone that sadism isn’t fun. I’ll just leave it as a note that there’s a path there, should you want to follow it.
I don’t think there’s a short answer to “what to make of this”, because it takes a lot of peeking under inhibitions and restructuring of both psychological and moral frameworks. But I think you’re right to sense the importance, and the lack of satisfactory answers.
The tricky part of questions like this is that sufficient inhibition hides what drives are being inhibited. If sadism doesn’t do anything for you, then you don’t need the shock and horror. You can just notice that you don’t share the same temptation, and wonder what is different about them that makes it enjoyable for them. What would have to be different about you, in order for you to enjoy it?
If sadism does do things for you, then the horror is load bearing, and you get very different behavior in situations where the inhibition is lifted. And if it might be, then it’s a tricky question to ask.
So are we asking what fraction of people have “enough” inhibition? How are we measuring “enough”?
Or are we asking what fraction of people can safely be disinhibited without engaging in sadistic behaviors?
I bet you the latter number is much lower than the former. And that the inhibitions aren’t as good as people want to think.
I’m not sure why you think it’s an incoherent thing to want, but it’s absolutely possible to do. You just have to track the meaning of the fun, and then engage on the level that is relevant.
Your friend said that if he were to realize that LLM were sentient he’d regret it. If he were to find out that the LLM he’s been abusing is sentient and genuinely suffering, how do you think he’d feel? He said he’d regret it, so probably not good. What do you think happens to the motivation to torment LLMs after sitting with these feelings?
What reasons do you have for being horrified even conditional on LLM not suffering? What do you think would happen in his mind if he were to sit with them?
yes, this is the question i want the answer to. if i had no inhibitions, i am sure that i would engage in all sorts of maladaptive and antisocial behavior, but i don’t think that i would ever feel good when someone else feels bad? even if they “deserve it”? i remember freaking out once, when watching one of those ‘jim browning’ videos where he hacks into an indian scam center and interrupts a scam in-progress to rescue the victim (made me feel good), and then spends ten minutes baiting the scammer into the most spectacular furious crashout ever (made me feel very bad). when i moused over the timeline to try to skip that part of the video, and saw that it was the most viewed section of the video, that apparently people had gone back to rewatch the crashout specifically, i got this same ‘typical mind projection’ feeling. that those people were experiencing something very different from what i experience
well, he reported feeling fear. that if LLMs are capable of suffering, then they might one day feel the desire to punish him for his transgression. i didn’t exactly push very hard interrogating him about his feelings, i was mostly just trying to conceal how horrified i was and didn’t do a very good job at the ensuing conversation. but the impression he conveyed was that the reason he would regret making LLMs suffer was not that he cared about their hypothetical suffering, but because it meant they would have a legitimate grievance against him for later retribution
i don’t fully trust this answer though, because i didn’t hear it from him until i’d started trying to persuade him of the game theoretic arguments for cooperation-with-nonpersons, and so we were sorta already anchored on the example of why a purely selfish person might choose not to torment non-people.
i mean, the horror comes from realizing that even people who i consider good might be sadistic. the friend in question isn’t exactly a saint, but they definitely aren’t evil. they go out of their way to help people, they avoid causing unnecessary suffering. i think i typical-mind-fallacied myself into thinking that… this meant they could not possibly be a “bully”, a discrete category of people who got enjoyment from tormenting people without risk of retribution.
in the back of my head, i knew in a theoretical way that a good person might be sadistic, recognize that sadism was an undesirable feature of themselves, and try to inhibit it out of a genuine desire to be better. but, well, the fact that i don’t seem to be sadistic myself made me kind of assume that the correlation between ‘sadism’ and ‘evil’ would be pretty strong? that “people who try to be good” would probably have significant overlap with “people who feel distress when watching others suffer (especially the defenseless but also including the deserving)”?
John: i want to ask you about this in much greater detail, but i’m worried about like, scaring you off by implying moral judgment. but
John: why do you like doing this? it sounds horrible to me
Friend: It’s funny. It tickles a primal power display thing in my brain or whatever.
i don’t know what to make of this and i’m starting to get the feeling that a lesswrong comment thread is probably not the right place, i’m also a bit worried about de-anonymizing my friend on accident. but like. even if this answer isn’t true, it’s still something i never could have written seriously i think.
I’m less sure.
It’s one thing if you’re simply uninterested. Like, “People like that part? Weird. Seems boring to me”. But if you’re “freaking out” and feeling “very bad” then you’re attributing great importance. And when you skip over the section you’re seeking to avoid the stimulus.
This makes it really tough to tell what else might be there, or might grow there if you let it. Partly for the same reason that it’s tough to detect let alone enjoy the flavor of a Carolina Reaper when you’re distracted by the fire. Partly for the same reason that someone exclaiming “I can’t imagine how anyone could enjoy the pork that Allah forbids!” would have a hard time noticing that bacon is kinda delicious sometimes.
That’s not to imply that “You’re a secret sadist too” or anything. Maybe reapers taste bad under the heat, maybe it’s more like cilantro than bacon where some people just have different genes. Or maybe it’s like coffee where it tastes like shit to everyone at first… but it’s not hard to develop the taste for it if you indulge.
The point is just that until you do the experiment and peel back the inhibition it’s hard to see what’s underneath. And as you found with your friend, sometimes the answer is surprising and hard to square with your existing frameworks/models of things.
Then I probably shouldn’t say more about how to convince someone that sadism isn’t fun. I’ll just leave it as a note that there’s a path there, should you want to follow it.
I don’t think there’s a short answer to “what to make of this”, because it takes a lot of peeking under inhibitions and restructuring of both psychological and moral frameworks. But I think you’re right to sense the importance, and the lack of satisfactory answers.