LESSWRONG
LW

0

Using lying to detect human values

by Stuart_Armstrong

15th Mar 2018

AI Alignment Forum

1 min read

0

This is a linkpost for https://www.lesserwrong.com/posts/pQz97SLCRMwHs6BzF/using-lying-to-detect-human-values

0

New Comment

More from Stuart_Armstrong

79Go home GPT-4o, you’re drunk: emergent misalignment as lowered inhibitions

Stuart_Armstrong, rgorman

1mo

12

170Using GPT-Eliezer against ChatGPT Jailbreaking

Stuart_Armstrong, rgorman

2y

85

67Alignment can improve generalisation through more robustly doing what a human wants - CoinRun example

Stuart_Armstrong

1y

9

Curated and popular this week

282Tracing the Thoughts of a Large Language Model

2d

24

191Impact, agency, and taste

7d

10

137Early Chinese Language Media Coverage of the AI 2027 Report: A Qualitative Analysis

1d

4