Pedro Domingos

Hm, don't think you quite understood my analogy. Let me rephrase: What is political about not doing research that, as part of its methodology, makes ppl sick or puts their lives at risk, without their consent? Safety & harm can be separate from ideology. Esp when data is involved

You're shifting the goalposts. The NeurIPS ethics review covers a lot more than physically harming people, and applies to things it makes no sense to apply it to.

So then we agree that ethics reviews make sense, the only question is what should be covered in an ethics review, yes?

Are you deliberately trying to misconstrue what I say, or is it involuntary?

“The NeurIPS ethics review covers a lot more than physically harming people...” This makes it sound like you object to the current scope of the ethics review, but you might consider one focused on physical harms more reasonable. Do I misunderstand?

No, I object to the ethics review process in toto, as I've made abundantly clear, and I point out yet again that NeurIPS is the outlier in this respect.

Neurips isn't an outlier. ACL has an ethics review, ACM is working on it. Other CS disciplines, such as security, take it seriously. In fact, the only outlier here is the ML/AI field for its stubbornness to get on board with something being done by by so many other fields.

That statement is factually false, and you know it.