lumenwrites

[Question] I want to donate some money (not much, just what I can af­ford) to AGI Align­ment re­search, to what­ever or­ga­ni­za­tion has the best chance of mak­ing sure that AGI goes well and doesn’t kill us all. What are my best op­tions, where can I make the most differ­ence per dol­lar?

What are these “out­side of the Over­ton win­dow” ap­proaches to pre­vent­ing AI apoc­a­lypse that Eliezer was talk­ing about in his post?

[Question] How would you ex­plain Bayesian think­ing to a ten year old?

• lumenwrites 29 Dec 2021 14:02 UTC

  8 points

  on: Book Launch: The Eng­ines of Cognition

  Looks amazing!

  I’d love to buy an ebook version though. Or even better—an audiobook.

• lumenwrites 1 Dec 2021 17:52 UTC

  1 point

  in reply to: Mitchell_Porter’s comment on: The Ra­tion­al­ists of the 1950s (and be­fore) also called them­selves “Ra­tion­al­ists”

  I don’t understand, its factors are 101 and 109, both are more than 100.

• lumenwrites 1 Dec 2021 9:36 UTC

  5 points

  in reply to: Shmi’s comment on: How do you write origi­nal ra­tio­nal­ist es­says?

  While I’d rather not test this empirically, I think I’m feeling pretty motivated to do this, and yet I can’t. I’d really like to solve this issue without resorting to hiring a professional assassin on myself.

[Question] How do you write origi­nal ra­tio­nal­ist es­says?

• lumenwrites 1 Dec 2021 7:52 UTC

  14 points

  on: The Ra­tion­al­ists of the 1950s (and be­fore) also called them­selves “Ra­tion­al­ists”

  That poem was amazing.

  How does a person factorize 11,009 in their head?

• lumenwrites 4 Apr 2021 6:23 UTC

  10 points

  on: Pre­dic­tive Cod­ing has been Unified with Backpropagation

  You guys will probably find this Slate Star Codex post interesting:

  https://​​slatestarcodex.com/​​2017/​​09/​​05/​​book-review-surfing-uncertainty/​​

  Scott summarizes the Predictive Processing theory, explains it in a very accessible way (no math required), and uses it to explain a whole bunch of mental phenomena (attention, imagination, motor behavior, autism, schizophrenia, etc.)

  Can someone ELI5/​TLDR this paper for me, explain in a way more accessible to a non-technical person?

  - How does backprop work if the information can’t flow backwards?
  - In Scotts post, he says that when lower-level sense data contradicts high-level predictions, high-level layers can override lower-level predictions without you noticing it. But if low-level sensed data has high confidence/​precision—the higher levels notice it and you experience “surprise”. Which one of those is equivalent to the backdrop error? Is it low-level predictions being overridden, or high-level layers noticing the surprise, or something else, like changing the connections between neurons to train the network and learn from the error somehow?