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.

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

94moShow Highlight
1

Clarifying "AI Alignment"

154moShow Highlight
41

An unaligned benchmark

84moShow Highlight
0

Prosaic AI alignment

104moShow 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

34moShow Highlight
1

Approval-directed bootstrapping

44moShow Highlight
0

Humans Consulting HCH

44moShow Highlight
0

Corrigibility

94moShow 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

54moShow Highlight
6

Benign model-free RL

34moShow Highlight
0

Factored Cognition

104moShow Highlight
1
0

AlphaGo Zero and capability amplification

62moShow 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

102moShow Highlight
1

The reward engineering problem

32moShow Highlight
1

Capability amplification

72moShow Highlight
1

Learning with catastrophes

72moShow Highlight
1

Possible approaches

The fifth section of the sequence breaks down some of these problems further and describes some possible approaches.

Thoughts on reward engineering

52moShow Highlight
12
1

Reliability amplification

42moShow Highlight
1