6 Matching Annotations
  1. Feb 2021
    1. Koo's discovery makes it possible to peek inside the black box and identify some key features that lead to the computer's decision-making process.

      Moving towards "explainable AI".

    2. Neural nets learn and make decisions independently of their human programmers. Researchers refer to this hidden process as a "black box." It is hard to trust the machine's outputs if we don't know what is happening in the box.

      Counter-argument: Why do we trust a human being's decisions if we don't know what is happening inside their brain? Yes, we can question the human being but we then have to trust that what they tell us about their rationale is true.

    1. We've developed scientific methods to study black boxes for hundreds of years now, but these methods have primarily been applied to [living beings] up to this point

      It's called psychology.

  2. Aug 2020
  3. Dec 2019
    1. Second, there are some fantastic projects out there like Create React App, preact-cli, nwb, and much more that avoid the boilerplate problem but at the expense of some other tradeoffs. Your configuration could be black-boxed and not able to be modified. They could force you to eject your configuration, making maintenance of the entire build dependency tree and configuration your responsibility again, and also preclude future configuration updates.
  4. Jan 2017
    1. An anomaly is thus a mere difference in degree for which the norm will serve as metric.

      Normal is a powerful, potent, and potentially pernicious black box. And it is frequently a black box deployed against rhetoric.