AI ALIGNMENT FORUM
Wikitags
AF

Subscribe
Discussion0
1
Deconfusion
Abram Demski

Deconfusion

Subscribe
Discussion0
1
Written by Abram Demski last updated 17th Mar 2021

Summaries

Narrowly, deconfusion is a specific branch of AI alignment research, discussed in MIRI's 2018 research update. More broadly, the term applies to any domain. Quoting from the research update:

By deconfusion, I mean something like “making it so that you can think about a given topic without continuously accidentally spouting nonsense.”

Posts tagged Deconfusion
12
26Looking Deeper at Deconfusion
Adam Shimi
4y
2
8
44Builder/Breaker for Deconfusion
Abram Demski
3y
8
8
16Traps of Formalization in Deconfusion
Adam Shimi
4y
2
2
81. A Sense of Fairness: Deconfusing Ethics
Roger Dearnaley
1y
0
1
42Deconfusing Direct vs Amortised Optimization
Beren Millidge
2y
3
1
35Modelling Transformative AI Risks (MTAIR) Project: Introduction
David Manheim, Aryeh Englander
4y
0
1
25Applications for Deconfusing Goal-Directedness
Adam Shimi
4y
3
1
21Musings on general systems alignment
Alex Flint
4y
1
1
15Open problem: how can we quantify player alignment in 2x2 normal-form games?
Q
Alex Turner, Vanessa Kosoy
4y
Q
32
1
12A review of "Agents and Devices"
Adam Shimi
4y
0
1
11Goal-Directedness and Behavior, Redux
Adam Shimi
4y
2
1
10Approaches to gradient hacking
Adam Shimi
4y
7
1
12Alex Turner's Research, Comprehensive Information Gathering
Adam Shimi
4y
3
1
10Power-seeking for successive choices
Adam Shimi
4y
9
1
137Simulators
janus
3y
90
Load More (15/21)
Add Posts