Parameter counts in Machine Learning

Jsevillamol19 Jun 2021 16:04 UTC

31 points

5 comments7 min readLW link

Knowledge is not just precipitation of action

alexflint18 Jun 2021 23:26 UTC

15 points

3 comments7 min readLW link

Non-poisonous cake: anthropic updates are normal

Stuart_Armstrong18 Jun 2021 14:51 UTC

22 points

4 comments1 min readLW link

[Question] Pros and cons of working on near-term technical AI safety and assurance

alenglander17 Jun 2021 20:17 UTC

11 points

1 comment2 min readLW link

[AN #152]: How we’ve overestimated few-shot learning capabilities

rohinmshah16 Jun 2021 17:20 UTC

20 points

6 comments8 min readLW link

(mailchi.mp)

Reward Is Not Enough

Steven Byrnes16 Jun 2021 13:52 UTC

76 points

14 comments10 min readLW link

[Question] Open problem: how can we quantify player alignment in 2x2 normal-form games?

TurnTrout16 Jun 2021 2:09 UTC

21 points

54 comments1 min readLW link

Vignettes Workshop (AI Impacts)

Daniel Kokotajlo15 Jun 2021 12:05 UTC

44 points

1 comment1 min readLW link

Knowledge is not just digital abstraction layers

alexflint15 Jun 2021 3:49 UTC

16 points

4 comments5 min readLW link

Looking Deeper at Deconfusion

adamShimi13 Jun 2021 21:29 UTC

39 points

11 comments15 min readLW link

Avoiding the instrumental policy by hiding information about humans

paulfchristiano13 Jun 2021 20:00 UTC

31 points

2 comments2 min readLW link

Answering questions honestly given world-model mismatches

paulfchristiano13 Jun 2021 18:00 UTC

29 points

1 comment16 min readLW link

(ai-alignment.com)

Finite Factored Sets: Orthogonality and Time

Scott Garrabrant10 Jun 2021 1:22 UTC

31 points

1 comment4 min readLW link

Knowledge is not just mutual information

alexflint10 Jun 2021 1:01 UTC

16 points

2 comments4 min readLW link

A naive alignment strategy and optimism about generalization

paulfchristiano10 Jun 2021 0:10 UTC

41 points

3 comments3 min readLW link

(ai-alignment.com)

“Decision Transformer” (Tool AIs are secret Agent AIs)

gwern9 Jun 2021 1:06 UTC

34 points

4 comments1 min readLW link

(sites.google.com)

AXRP Episode 8 - Assistance Games with Dylan Hadfield-Menell

DanielFilan8 Jun 2021 23:20 UTC

9 points

0 comments71 min readLW link

The Inside View #3: Evan Hubinger— homogeneity in takeoff speeds, learned optimization and interpretability

Michaël Trazzi8 Jun 2021 19:20 UTC

28 points

0 comments55 min readLW link

Survey on AI existential risk scenarios

Sam Clarke, Alexis Carlier and Jonas Schuett

8 Jun 2021 17:12 UTC

54 points

10 comments7 min readLW link

The reverse Goodhart problem

Stuart_Armstrong8 Jun 2021 15:48 UTC

14 points

22 comments1 min readLW link