AI ALIGNMENT FORUM
AF

"Why Not Just..."

Aug 08, 2022 by johnswentworth

A compendium of rants about alignment proposals, of varying charitability.

52Deep Learning Systems Are Not Less Interpretable Than Logic/Probability/Etc
johnswentworth
3y
14
47Godzilla Strategies
johnswentworth
3y
19
43Rant on Problem Factorization for Alignment
johnswentworth
3y
32
59Interpretability/Tool-ness/Alignment/Corrigibility are not Composable
johnswentworth
3y
4
77How To Go From Interpretability To Alignment: Just Retarget The Search
johnswentworth
3y
24
38Oversight Misses 100% of Thoughts The AI Does Not Think
johnswentworth
3y
23
37Human Mimicry Mainly Works When We’re Already Close
johnswentworth
3y
4
66Worlds Where Iterative Design Fails
johnswentworth
3y
17
69Why Not Just... Build Weak AI Tools For AI Alignment Research?
johnswentworth
2y
2
46Why Not Just Outsource Alignment Research To An AI?
johnswentworth
2y
7
48OpenAI Launches Superalignment Taskforce
Zvi
2y
0