Iterated Amplification

Oct 29, 2018

by paulfchristiano

This is an upcoming sequence curated by Paul Christiano on one current approach to alignment: Iterated Amplification. The posts will be released in the coming weeks, on average every 2-3 days.

Preface to the sequence on iterated amplification

122mo2 min readShow Highlight
0

Problem statement

The first part of this sequence clarifies the problem that iterated amplification is trying to solve, which is both narrower and broader than you might expect.

The Steering Problem

92mo7 min readShow Highlight
1

Clarifying "AI Alignment"

152mo3 min readShow Highlight
41

An unaligned benchmark

82mo9 min readShow Highlight
0

Prosaic AI alignment

102mo7 min readShow Highlight
0

Basic intuition

The second part of the sequence outlines the basic intuitions that motivate iterated amplification. I think that these intuitions may be more important than the scheme itself, but they are considerably more informal.

Approval-directed agents: overview

32mo8 min readShow Highlight
1

Approval-directed agents: details

42mo7 min readShow Highlight
0

Approval-directed bootstrapping

42mo1 min readShow Highlight
0

Humans Consulting HCH

42mo1 min readShow Highlight
0

Corrigibility

92mo6 min readShow Highlight
2

The scheme

The core of the sequence is the third section. Benign model-free RL describes iterated amplification, as a general outline into which we can substitute arbitrary algorithms for reward learning, amplification, and robustness. The first four posts all describe variants of this idea from different perspectives, and if you find that one of those descriptions is clearest for you then I recommend focusing on that one and skimming the others.

Iterated Distillation and Amplification

52mo6 min readShow Highlight
6

Benign model-free RL

32mo7 min readShow Highlight
0

Factored Cognition

102mo16 min readShow Highlight
1

[Link] Supervising strong learners by amplifying weak experts

717d1 min readShow Highlight
0

AlphaGo Zero and capability amplification

615d2 min readShow Highlight
16

What needs doing

The fourth part of the sequence describes some of the black boxes in iterated amplification and discusses what we would need to do to fill in those boxes. I think these are some of the most important open questions in AI alignment.

Directions and desiderata for AI alignment

1010d13 min readShow Highlight
1

The reward engineering problem

37d6 min readShow Highlight
1

Capability amplification

73d13 min readShow Highlight
1

Learning with catastrophes

515h3 min readShow Highlight
0