3 ms·

If a community has a stance on US politics which interact with it only incidentally — which the community very much does have, to the point of distraction, it i

If a community has a stance on US politics which interact with it only incidentally — which the community very much does have, to the point of distraction, it is alleged — then the community can damn well afford to have a stance on international politics when the advancements the community builds are key tools in building the engines of tyrannical repression.

LatteLazy 6y ago
I think it's fine to take a stance on things central to the groups main function: don't supply ML Tech for immoral use. But beyond that? That's a lot of issues you need to make stances on before you can actually do any machine learning work...
- SpicyLemonZest 6y ago
  This issue isn't beyond that. The question is how a commitment to not supply ML tech for immoral use is compatible with supplying it to a country that's genociding the Uighurs.
- crispyambulance 6y ago
  > don't supply ML Tech for immoral use. But beyond that? There's a lot to that and it goes beyond the scope of what researchers can handle by themselves. It's an effort that involves more than just the researchers, though they will take a central role. For one thing, how do you evaluate whether a usage is "immoral"? How do you enforce correct usage? These are difficult questions with multifaceted answers which no one has yet elucidated (yes, it has to be more than "I'll know it when I see it").
  - LatteLazy 6y ago
    This is exactly my point. You can get general agreement to "do no evil". But every step you take to be specific makes the amount of discussion and detail exponentially more difficult. So you can't have a specific stance on everything. You either take a specific stance on a small number of issues central to the community or a general stance, or you spend all your time creating and amending stances and do nothing else.

3 ms·

If a community has a stance on US politics which interact with it only incidentally — which the community very much does have, to the point of distraction, it i

by fennecfoxen 6y ago

LatteLazy 6y ago
I think it's fine to take a stance on things central to the groups main function: don't supply ML Tech for immoral use. But beyond that? That's a lot of issues you need to make stances on before you can actually do any machine learning work...
- SpicyLemonZest 6y ago
  This issue isn't beyond that. The question is how a commitment to not supply ML tech for immoral use is compatible with supplying it to a country that's genociding the Uighurs.
- crispyambulance 6y ago
  > don't supply ML Tech for immoral use. But beyond that? There's a lot to that and it goes beyond the scope of what researchers can handle by themselves. It's an effort that involves more than just the researchers, though they will take a central role. For one thing, how do you evaluate whether a usage is "immoral"? How do you enforce correct usage? These are difficult questions with multifaceted answers which no one has yet elucidated (yes, it has to be more than "I'll know it when I see it").
  - LatteLazy 6y ago
    This is exactly my point. You can get general agreement to "do no evil". But every step you take to be specific makes the amount of discussion and detail exponentially more difficult. So you can't have a specific stance on everything. You either take a specific stance on a small number of issues central to the community or a general stance, or you spend all your time creating and amending stances and do nothing else.