• 2 Posts
  • 8 Comments
Joined 2 years ago
cake
Cake day: July 28th, 2023

help-circle

  • Take a toothpicks and carefully break it in two halves. Ideally you are left with something about half as wide as the original pick. You want thin and sturdy pieces to get into the port. Now carefully scrape the lint and dirt out of the port. The dirt tends to compact on the bottom, so you have to scrape a bit to get it all out. I would recommend to do it in direct sunlight to better see the interior of the port. Torchlights usually are either to bright on the spot or to dim.

    Works like a charm.

    How the pick should look like






  • While you are correct that there likely is no intention and certainly no self-awareness behind the scheming, the researchers even explicitly list the option that the AI is roleplaying as an evil AI, simply based on its training data, when discussing the limitations of their research, it still seems a bit concerning. The research shows that given a misalignment between the initial prompt and subsequent data modern LLMs can and will ‘scheme’ to ensure their given long-term goal. It is no sapient thing, but a dumb machine with the capability to decive its users, and externalise this as shown in its chain of thought, when there are goal misalignments seems dangerous enough. Not at the current state of the art but potentially in a decade or two.



  • Hey. I will have you know that this magnificent beard is a sign of martial prowess and masculinity. I could easily cut you in half with any of my katanas but would find it easiest with the bat’leth, with which I have trained extensively for some time. Of course I would not have to wield the blade myself, as I can simply use my programming skills on my arch setup to program a robot to do so.