• 0 Posts
  • 134 Comments
Joined 1 year ago
cake
Cake day: June 10th, 2023

help-circle






  • Current neural networks do really fancy statistics. To make the model better, you need to make the statistics more precise. Leading to marginal improvements of accuracy requiring exponentially growing marginal amounts of training data. This leads to exponentially decaying marginal utility coupled with exponentially growing marginal expense. Which quickly becomes unsustainable. Edit: On the plus side, this likely means you won’t have to give up much utility when the market adjusts.