Chronotope’s avatarChronotope’s Twitter Archive—№ 108,425

        1. …in reply to @DrMichaelAnn
          smarmyliberal @veronicaripson swodinsky I mean, not give these systems literally any slack, but also they collect plenty of user data already and selected targeting methodologies is likely not the place they are doing a ton of training ML on, because marketers, generally, are real bad at targeting assumptions.
      1. …in reply to @Chronotope
        smarmyliberal @veronicaripson swodinsky I mean, I'm sure the systems themselves assume they are far more accurate then they are, based on the ML they do on user data. That accuracy is def a sacrifice to the training sets.
    1. …in reply to @Chronotope
      smarmyliberal @veronicaripson swodinsky But while lots of stupid people are doing stupid shit in Silicon Valley, For Sure, that would just be a super terrible training set and I really hope no one is even thinking about using it. Ex: howtonotsuckatgamedesign.com/2013/12/marketers-fear-female-geek-2/
  1. …in reply to @Chronotope
    smarmyliberal @veronicaripson swodinsky *that - being what marketers are choosing to target.


Search tweets' text