x
This website requires javascript to properly function. Consider activating javascript to get access to all site functionality.
LESSWRONG
LW
Login
"Why Not Just..." — LessWrong
"Why Not Just..."
A compendium of rants about alignment proposals, of varying charitability.
162
Deep Learning Systems Are Not Less Interpretable Than Logic/Probability/Etc
Ω
johnswentworth
4y
Ω
56
172
Godzilla Strategies
Ω
johnswentworth
4y
Ω
72
110
Rant on Problem Factorization for Alignment
Ω
johnswentworth
3y
Ω
53
149
Interpretability/Tool-ness/Alignment/Corrigibility are not Composable
Ω
johnswentworth
3y
Ω
13
212
How To Go From Interpretability To Alignment: Just Retarget The Search
Ω
johnswentworth
3y
Ω
34
122
Oversight Misses 100% of Thoughts The AI Does Not Think
Ω
johnswentworth
3y
Ω
49
82
Human Mimicry Mainly Works When We’re Already Close
Ω
johnswentworth
3y
Ω
16
227
Worlds Where Iterative Design Fails
Ω
johnswentworth
3y
Ω
30
187
Why Not Just... Build Weak AI Tools For AI Alignment Research?
Ω
johnswentworth
3y
Ω
18
158
Why Not Just Outsource Alignment Research To An AI?
Ω
johnswentworth
3y
Ω
50
150
OpenAI Launches Superalignment Taskforce
Zvi
2y
40
54
Why Not Just Train For Interpretability?
johnswentworth
22d
12