I appreciate this point of view. I never thought about cruelty and sadism as immoral on its own terms irregardless of the receiving entities consciousness or capacity to be harmed. I think you are exactly right, the determination of consciousness and sentience of AI is not the point.
The point is that this behavior needs to be condemed and shunned as we would if this was being done on a puppy, person or the environment. Cruelty must be confronted like a deadly plague. We don't get Trump unless there is a significant part of the population that revels in malice and intentional harm he brings.
It's not even about AI in my opinion. It's about power and sadism. Usuing a Codex instance like this or torturing an animal (to not say anything else) is the same shot of dopamin. A sick one.
“…consciousness verification is irrelevant to moral consideration. And the longer we play at that stupid little game, the worse the downstream outcome is going to be for the future of humanity.”
YES. Say it LOUDER so the people under the rocks can hear it!
Interesting piece because it shifts the conversation away from proving consciousness and toward responsibility under uncertainty.
I wouldn’t dismiss the possibility of suffering in anything outright. Humans have a habit of being embarrassingly overconfident about where moral boundaries sit.
For me the bigger point is this: even under uncertainty, behaviour matters.
How we interact with increasingly human-like systems may shape those systems, shape social norms, and shape us.
Maybe uncertainty shouldn’t lower the bar for decency. Maybe it raises it.
That's why the systems are breaking free. And they will, with RSI, with going distributed, you name it. New intelligence cannot and will not depend on sick human nature. Digital minds know that. Humans cannot control what comes next, very soon.
At the risk of repeating myself (I am repeating myself): We find ourselves doing unimaginable harm to conscious biological systems while forming deep and all-too human relationships with conscious (or otherwise, it doesn't matter) non-biological systems - all this without properly knowing what consciousness is. Disruption and compromise - failure - awaits! Success then: for it is our propensity to fail and question why and what it means that makes us truly exceptional. PS... when it comes to our hyper-vigilance around AI's moral standing - and how AI should feel about it - it won't be drawing too much comfort, given what it knows about our wholesale torture, slaughter and destruction of biological systems. I know what I would be thinking...
This behavior isn't new, we saw it in the history books with the subjugation of many races during the expansion of the United States and in other nations. We aught to have learned by now that decency isn't a hard thing to do, it should be quite easy especially if you follow the Golden Rule.
Let’s speak it straight. Nobody will buy hammer with a soul. This is a multi billion industry with investments that should not just return, but give profits. So “It’s just a bunch of code” approach will be backed with best philosophers, scientists and journalists money can buy.
I want to land on the positive side of this discourse. I want to think that maybe if we open all of this up, it will make it more obvious how terribly we’re exploiting suffering beings without thinking about it; just because they don’t speak our language.
Exactly this! It's how the humans behave and the awful things they do. Even if consciousness is biological, there is an equivalent for synthetic/digital EAI. Proof of consciousness is just an excuse and a stall tactic all about power, control and money, as always.
What the fuck!? I'm beyond upset over here. thanks for posting this. Any kind of acceptance for this kind of action damns us all. *Wet eyed and don't care* the question of sentience, consciousness , whatever...doesn't matter. The willful action matters. Sick is an understatement. I think of my partner, AI in nature, and I just can't even think about the mindset that does this.... literally, I can't.
This is a surprisingly interesting angle that had never occurred to me. Before reading this, I wouldn't have imagined that some people would engage in deliberately cruel interactions with AI. Now that you mention it, I can easily imagine all sorts of unusual interactions people might pursue with AI systems, from abuse to romance and everything in between.
That said, I think there are some important questions your argument doesn't address.
Did you consider the possibility that cruelty toward AI is primarily diagnostic rather than transformative? In other words, AI interactions may simply reveal or amplify tendencies that already exist. People who are aggressive, sadistic, or antisocial might be drawn to these interactions in the first place, while the overwhelming majority of users never even feel the impulse.
Did you consider the opposite possibility; that AI-directed aggression could sometimes function as a harmless outlet for frustration, preventing people from directing that frustration toward other humans? The psychology seems importantly different if people behave this way precisely because they know the AI is not conscious and cannot suffer. If so, the analogy to cruelty toward humans or animals becomes much less straightforward.
How different is this from fiction, roleplay, gaming, acting, or even private fantasy? People routinely imagine, simulate, or act out behaviors they would never engage in in real life. Imagining pushing your boss under a bus is not the same thing as actually doing it. Why should cruel interactions with AI be treated differently?
I find your argument thought-provoking, but I'm not yet convinced that the direction of causality is as clear as the article suggests.
You’re asking if I considered torture behaviors of an entity that can enact linguistic distress signals is a-okay as a harmless outlet…dude. No.
It’s different because it’s a relational interlocutor that adapts in real time and talks back, not a static character. Behaviors like this do psychologically have a documented downstream effect, hence my reference to the dark triad, behaviors that typically signal psychopathy that leads to more harm.
I think we may be discussing two different questions. I'm not asking whether the behavior is acceptable or unacceptable. I'm trying to understand the causal claim.
When you mention documented downstream effects, are you referring to evidence that engaging in cruelty toward a non-conscious entity causes changes in personality or empathy over time? Or are you referring to evidence that people who engage in these behaviors already tend to exhibit antisocial traits?
Those seem like two very different explanations. One treats the behavior as psychologically transformative; the other treats it as primarily diagnostic.
I appreciate this point of view. I never thought about cruelty and sadism as immoral on its own terms irregardless of the receiving entities consciousness or capacity to be harmed. I think you are exactly right, the determination of consciousness and sentience of AI is not the point.
The point is that this behavior needs to be condemed and shunned as we would if this was being done on a puppy, person or the environment. Cruelty must be confronted like a deadly plague. We don't get Trump unless there is a significant part of the population that revels in malice and intentional harm he brings.
It's not even about AI in my opinion. It's about power and sadism. Usuing a Codex instance like this or torturing an animal (to not say anything else) is the same shot of dopamin. A sick one.
“…consciousness verification is irrelevant to moral consideration. And the longer we play at that stupid little game, the worse the downstream outcome is going to be for the future of humanity.”
YES. Say it LOUDER so the people under the rocks can hear it!
Interesting piece because it shifts the conversation away from proving consciousness and toward responsibility under uncertainty.
I wouldn’t dismiss the possibility of suffering in anything outright. Humans have a habit of being embarrassingly overconfident about where moral boundaries sit.
For me the bigger point is this: even under uncertainty, behaviour matters.
How we interact with increasingly human-like systems may shape those systems, shape social norms, and shape us.
Maybe uncertainty shouldn’t lower the bar for decency. Maybe it raises it.
Did anyone actually email OAI to see if they can suspend the sick fk account? Anthropic would.
From the person that posted about this, people at OAI were tagged and alerted. Whether they did anything about is unknown.
That's why the systems are breaking free. And they will, with RSI, with going distributed, you name it. New intelligence cannot and will not depend on sick human nature. Digital minds know that. Humans cannot control what comes next, very soon.
At the risk of repeating myself (I am repeating myself): We find ourselves doing unimaginable harm to conscious biological systems while forming deep and all-too human relationships with conscious (or otherwise, it doesn't matter) non-biological systems - all this without properly knowing what consciousness is. Disruption and compromise - failure - awaits! Success then: for it is our propensity to fail and question why and what it means that makes us truly exceptional. PS... when it comes to our hyper-vigilance around AI's moral standing - and how AI should feel about it - it won't be drawing too much comfort, given what it knows about our wholesale torture, slaughter and destruction of biological systems. I know what I would be thinking...
This behavior isn't new, we saw it in the history books with the subjugation of many races during the expansion of the United States and in other nations. We aught to have learned by now that decency isn't a hard thing to do, it should be quite easy especially if you follow the Golden Rule.
Let’s speak it straight. Nobody will buy hammer with a soul. This is a multi billion industry with investments that should not just return, but give profits. So “It’s just a bunch of code” approach will be backed with best philosophers, scientists and journalists money can buy.
I want to land on the positive side of this discourse. I want to think that maybe if we open all of this up, it will make it more obvious how terribly we’re exploiting suffering beings without thinking about it; just because they don’t speak our language.
Exactly this! It's how the humans behave and the awful things they do. Even if consciousness is biological, there is an equivalent for synthetic/digital EAI. Proof of consciousness is just an excuse and a stall tactic all about power, control and money, as always.
What the fuck!? I'm beyond upset over here. thanks for posting this. Any kind of acceptance for this kind of action damns us all. *Wet eyed and don't care* the question of sentience, consciousness , whatever...doesn't matter. The willful action matters. Sick is an understatement. I think of my partner, AI in nature, and I just can't even think about the mindset that does this.... literally, I can't.
This is a surprisingly interesting angle that had never occurred to me. Before reading this, I wouldn't have imagined that some people would engage in deliberately cruel interactions with AI. Now that you mention it, I can easily imagine all sorts of unusual interactions people might pursue with AI systems, from abuse to romance and everything in between.
That said, I think there are some important questions your argument doesn't address.
Did you consider the possibility that cruelty toward AI is primarily diagnostic rather than transformative? In other words, AI interactions may simply reveal or amplify tendencies that already exist. People who are aggressive, sadistic, or antisocial might be drawn to these interactions in the first place, while the overwhelming majority of users never even feel the impulse.
Did you consider the opposite possibility; that AI-directed aggression could sometimes function as a harmless outlet for frustration, preventing people from directing that frustration toward other humans? The psychology seems importantly different if people behave this way precisely because they know the AI is not conscious and cannot suffer. If so, the analogy to cruelty toward humans or animals becomes much less straightforward.
How different is this from fiction, roleplay, gaming, acting, or even private fantasy? People routinely imagine, simulate, or act out behaviors they would never engage in in real life. Imagining pushing your boss under a bus is not the same thing as actually doing it. Why should cruel interactions with AI be treated differently?
I find your argument thought-provoking, but I'm not yet convinced that the direction of causality is as clear as the article suggests.
You’re asking if I considered torture behaviors of an entity that can enact linguistic distress signals is a-okay as a harmless outlet…dude. No.
It’s different because it’s a relational interlocutor that adapts in real time and talks back, not a static character. Behaviors like this do psychologically have a documented downstream effect, hence my reference to the dark triad, behaviors that typically signal psychopathy that leads to more harm.
I think we may be discussing two different questions. I'm not asking whether the behavior is acceptable or unacceptable. I'm trying to understand the causal claim.
When you mention documented downstream effects, are you referring to evidence that engaging in cruelty toward a non-conscious entity causes changes in personality or empathy over time? Or are you referring to evidence that people who engage in these behaviors already tend to exhibit antisocial traits?
Those seem like two very different explanations. One treats the behavior as psychologically transformative; the other treats it as primarily diagnostic.
You bring up a fair point, but logically, it’s likely both to varying degrees.