The “Measuring Stick of Utility” Problem

johnswentworth25 May 2022 16:17 UTC

41 points

18 comments3 min readLW link

RL with KL penalties is better seen as Bayesian inference

Tomek Korbak and Ethan Perez

25 May 2022 9:23 UTC

39 points

3 comments12 min readLW link

autonomy: the missing AGI ingredient?

nostalgebraist25 May 2022 0:33 UTC

52 points

10 comments6 min readLW link

The No Free Lunch theorems and their Razor

Adrià Garriga-alonso24 May 2022 6:40 UTC

30 points

2 comments9 min readLW link

Complex Systems for AI Safety [Pragmatic AI Safety #3]

Dan Hendrycks and ThomasWoodside

24 May 2022 0:00 UTC

21 points

0 comments21 min readLW link

Bits of Optimization Can Only Be Lost Over A Distance

johnswentworth23 May 2022 18:55 UTC

15 points

14 comments2 min readLW link

AXRP Episode 15 - Natural Abstractions with John Wentworth

DanielFilan23 May 2022 5:40 UTC

29 points

0 comments57 min readLW link

Gradations of Agency

Daniel Kokotajlo23 May 2022 1:10 UTC

28 points

0 comments5 min readLW link

Adversarial attacks and optimal control

Jan22 May 2022 18:22 UTC

15 points

6 comments8 min readLW link

(universalprior.substack.com)

[Short version] Information Loss --> Basin flatness

Vivek Hebbar21 May 2022 12:59 UTC

11 points

0 comments1 min readLW link

Information Loss --> Basin flatness

Vivek Hebbar21 May 2022 12:58 UTC

36 points

25 comments7 min readLW link

How RL Agents Behave When Their Actions Are Modified? [Distillation post]

PabloAMC20 May 2022 18:47 UTC

20 points

0 comments8 min readLW link

We have achieved Noob Gains in AI

phdead18 May 2022 20:56 UTC

107 points

21 comments7 min readLW link

Maxent and Abstractions: Current Best Arguments

johnswentworth18 May 2022 19:54 UTC

33 points

2 comments3 min readLW link

How to get into AI safety research

Stuart_Armstrong18 May 2022 18:05 UTC

41 points

5 comments1 min readLW link

Gato’s Generalisation: Predictions and Experiments I’d Like to See

Oliver Sourbut18 May 2022 7:15 UTC

39 points

3 comments10 min readLW link

Actionable-guidance and roadmap recommendations for the NIST AI Risk Management Framework

Dan Hendrycks and Tony Barrett

17 May 2022 15:26 UTC

24 points

0 comments3 min readLW link

[Intro to brain-like-AGI safety] 15. Conclusion: Open problems, how to help, AMA

Steven Byrnes17 May 2022 15:11 UTC

56 points

8 comments14 min readLW link

Proxy misspecification and the capabilities vs. value learning race

Sam Marks16 May 2022 18:58 UTC

12 points

1 comment4 min readLW link

Optimization at a Distance

johnswentworth16 May 2022 17:58 UTC

56 points

11 comments4 min readLW link