LESSWRONGTags
LW

Iterated Amplification

EditHistory
Discussion (0)
Help improve this page (1 flag)
EditHistory
Discussion (0)
Help improve this page (1 flag)
Iterated Amplification
Random Tag
Contributors
2Ben Pace
2jacobjacob

Iterated Amplification is an approach to AI alignment, spearheaded by Paul Christiano. In this setup, we build powerful, aligned ML systems through a process of initially building weak aligned AIs, and recursively using each new AI to build a slightly smarter and still aligned AI. 

See also: Factored cognition. 

Posts tagged Iterated Amplification
11
46Iterated Distillation and AmplificationΩ
Ajeya Cotra
5y
Ω
14
9
126Paul's research agenda FAQΩ
zhukeepa
5y
Ω
74
9
120Challenges to Christiano’s capability amplification proposalΩ
Eliezer Yudkowsky
5y
Ω
54
8
70A guide to Iterated Amplification & DebateΩ
Rafael Harth
3y
Ω
9
5
15HCH and Adversarial Questions
David Udell
1y
7
4
33AlphaGo Zero and capability amplificationΩ
paulfchristiano
4y
Ω
23
3
127Debate update: Obfuscated arguments problemΩ
Beth Barnes
2y
Ω
23
3
119My Understanding of Paul Christiano's Iterated Amplification AI Safety Research AgendaΩ
Chi Nguyen
3y
Ω
20
2
206An overview of 11 proposals for building safe advanced AIΩ
evhub
3y
Ω
36
2
124My Overview of the AI Alignment Landscape: A Bird's Eye ViewΩ
Neel Nanda
2y
Ω
9
2
98Writeup: Progress on AI Safety via DebateΩ
Beth Barnes, paulfchristiano
3y
Ω
18
2
64Relaxed adversarial training for inner alignmentΩ
evhub
4y
Ω
27
2
60Garrabrant and Shah on human modeling in AGIΩ
Rob Bensinger
2y
Ω
10
2
60Prize for probable problemsΩ
paulfchristiano
5y
Ω
63
2
53CorrigibilityΩ
paulfchristiano
5y
Ω
8
Load More (15/60)
Add Posts