rusalkii

• rusalkii 23 Jul 2020 2:08 UTC

  23 points

  on: My Mat­ing Plan ala Ge­offrey Miller

  Undergrads and masters students are a promising market where I have high status. Starting a campus club or teaching GRE skills might help.

  Ignoring questions like tone or whether this post belongs on Less Wrong, this specifically is obviously unethical. Do not start a club to hit on undergrads who join it. Do not hit on your students. There is a power differential and you are on the wrong side of it.

• rusalkii 22 Feb 2023 21:51 UTC

  10 points

  4

  on: The male AI al­ign­ment solution

  I expect people are downvoting without explanation because, frankly, this reads like sufficiently obvious sexism it’s difficult to believe that the author hasn’t noticed. Assuming you want an actual explanation of what’s wrong with this post, I think there are two main parts:

  Epistemically speaking you are making very confident sweeping generalizations about something which is at best a tentative evopsych theory and at worst utter nonsense.

  Socially, this is incredibly dehumanizing and othering. Women are not alien intelligences. We think the same way you do. Ferreting out the fundamental intentions of men works the exact same way as ferreting out the fundamental intentions of women.

• rusalkii 25 Feb 2023 5:30 UTC

  1 point

  0

  in reply to: TekhneMakre’s comment on: The male AI al­ign­ment solution

  It would be somewhat less bad if it had been more clearly labeled speculative, but that’s not the fundamental issue. “cartoon” implies to me something like Newton’s laws—not correct exactly, but a good enough model to be going on for the purposes of the conversation. I think your object-level evopsych statements are closer to, uh, I don’t actually know physics nearly well enough to complete the analogy. Some sort of theory of a phenomenon that is not entirely proven to even exist, with some evidence for and some against, which a small minority group of scientists present as settled science and procede to write further papers using it as an assumption.

  I was not saying you had made the claim up, but presenting controversial claims with no hedging is not great. As for everything else, your post implies strongly, without stating outright, various narratives about human motivations/​evolution that are not, in fact, obvious. For instance, that women want to secure the loyalty of one man, while men want to have sex with as many women as possible, and that this adversarial dynamic is present in the modern day and results in women, in particular, having unique insight into figuring out the motives of partially aligned intelligences due to practice on men.

  It’s okay to describe features of a group of people. Which features you’re describing, how you present your claims, and whether you’re in fact right all matter. In this case, you are, implicitly, making the claim that the difference between men and women is large enough that it makes sense to try to draw an analogy to the difference between humans and AIs, even though you explicitly stated that of course the difference is not as large.

  To put it another way, I don’t actually see what using women and men here adds to the analogy beyond “sometimes, humans have to suss out the true intentions of other humans who partially share goals with them when those other humans have motive to deceive them”. To the extent that you are claiming there is a meaningful difference, I think that is [not entirely sure I am phrasing the following correctly] privileging gender as a special axis of human difference in a way that I think is meaningfully wrong and also find unpleasant.

  (Somewhat more incidentally, I and many other women I know dislike the use of “females”, “mate”, etc in this context, though that is somewhat trivial and not actually a big deal so much as often correlated with things that do actually bother me.)

• rusalkii 25 Feb 2023 5:34 UTC

  1 point

  0

  in reply to: Ruby’s comment on: The male AI al­ign­ment solution

  In addition to what gears said, I think the sexist othering etc is not actually critical to the analogy, which is kind of the problem. “Figuring out the motives of people who kind of share goals with you but also have reasons to lie” is a pretty universal human experience. Adding some gender evopsych on top is just annoying (and prevents thinking about many of the more interesting ways in which this dynamic can play out).