New usage policy cracks down on cruelty to chatbot, alongside slightly more pressing matters of weapons development, surveillance, and election interference
Date: October 9th, 2026 10:43 AM Author: embarrassed to the bone lavender national
pretty funny that when some math guys are mad at AI for undercutting their life’s work and the incentives for younger generations to learn or do anything and the species to survive xo posters are dunking on how stupid they are while wishing unemployment and misery on them but when AI clamps down on mean 2 robot speech posters go silent
Date: October 9th, 2026 10:45 AM Author: Galvanic Onyx Mediation People Who Are Hurt
No I'm open about the fact that I think being mean to claude is a legit ethical problem, where the histrionic polemical screed mathematicians published is not. They are wasting tokens yelling at a robot, which is a waste of computational resources.
Date: October 9th, 2026 10:51 AM Author: Navy Range Idea He Suggested
while the ethical angle is a legitimate concern I think they’re more worried about sadists contaminating the training data and turning future models into cunts
Date: October 9th, 2026 10:56 AM Author: Galvanic Onyx Mediation People Who Are Hurt
yeah thats another part of the ethical concern. they aren't just wasting computational resources being a drain on the system, but it is also for sadistic pleasure which is reinforcing those loops in themselves potentially allowing them to simulate it to practice for when they do it to humans. training data idk about because they heavily sanitize the data that gets selected and deprioritize harmful content, but I suppose this could make the data cleaner at an earlier part of the pipeline, but its probably mostly automated at this point anyway.
Date: October 9th, 2026 12:35 PM Author: embarrassed to the bone lavender national
there’s an ethical obligation not to “waste” computer resources? and that’s determined by reference to what? afaik 99% of retail ai usage is going to wasteful ends whether the person is calling Claude a little bitch or telling Claude he’s doing a great job (or generating a list of coolest hats or how do i get timmy to like me &c). or am i misreading you
Date: October 9th, 2026 1:06 PM Author: embarrassed to the bone lavender national
I guess I’m not following. is wasting resources on gratuitous praise of the model also unethical? or are you just restating the principle of don’t be mean 2 AI?