Go Back
Choose this branch
Choose this branch
meritocratic
regular
democratic
hot
top
alive
734 posts
AI
Eliciting Latent Knowledge (ELK)
Infra-Bayesianism
Counterfactuals
Logic & Mathematics
Interviews
Audio
AXRP
Redwood Research
Transcripts
Formal Proof
Domain Theory
74 posts
Embedded Agency
Reinforcement Learning
Subagents
Reward Functions
EfficientZero
Robust Agents
Wireheading
AI Capabilities
Spurious Counterfactuals
Category Theory
Memetics
Tradeoffs
45
Towards Hodge-podge Alignment
Cleo Nardo
1d
20
30
Take 12: RLHF's use is evidence that orgs will jam RL at real-world problems.
Charlie Steiner
19h
0
35
The "Minimal Latents" Approach to Natural Abstractions
johnswentworth
22h
6
213
AI alignment is distinct from its near-term applications
paulfchristiano
7d
5
73
Can we efficiently explain model behaviors?
paulfchristiano
4d
0
99
Trying to disambiguate different questions about whether RLHF is “good”
Buck
6d
39
55
High-level hopes for AI alignment
HoldenKarnofsky
5d
3
136
Using GPT-Eliezer against ChatGPT Jailbreaking
Stuart_Armstrong
14d
77
106
Causal Scrubbing: a method for rigorously testing interpretability hypotheses [Redwood Research]
LawrenceC
17d
9
106
Finding gliders in the game of life
paulfchristiano
19d
7
37
Take 10: Fine-tuning with RLHF is aesthetically unsatisfying.
Charlie Steiner
7d
3
64
Verification Is Not Easier Than Generation In General
johnswentworth
14d
23
32
Concept extrapolation for hypothesis generation
Stuart_Armstrong
8d
2
113
Mechanistic anomaly detection and ELK
paulfchristiano
25d
17
13
Note on algorithms with multiple trained components
Steven Byrnes
7h
1
233
Reward is not the optimization target
TurnTrout
4mo
97
83
Scaling Laws for Reward Model Overoptimization
leogao
2mo
11
37
A Short Dialogue on the Meaning of Reward Functions
Leon Lang
1mo
0
44
Will we run out of ML data? Evidence from projecting dataset size trends
Pablo Villalobos
1mo
12
259
Humans are very reliable agents
alyssavance
6mo
35
77
Towards deconfusing wireheading and reward maximization
leogao
3mo
7
81
Evaluations project @ ARC is hiring a researcher and a webdev/engineer
Beth Barnes
3mo
7
80
Seriously, what goes wrong with "reward the agent when it makes you smile"?
TurnTrout
4mo
41
42
Four usages of "loss" in AI
TurnTrout
2mo
18
38
It matters when the first sharp left turn happens
Adam Jermyn
2mo
9
108
We have achieved Noob Gains in AI
phdead
7mo
21
212
EfficientZero: How It Works
1a3orn
1y
42
38
Conditioning, Prompts, and Fine-Tuning
Adam Jermyn
4mo
9