LESSWRONG
LW

Concept Safety

Oct 02, 2019 by Kaj_Sotala

We can think of concepts and ontologies as tools for achieving our values. How does this influence our thinking on AI safety?

51Concept Safety: Producing similar AI-human concept spaces
Kaj_Sotala
10y
45
28Concept Safety: The problem of alien concepts
Kaj_Sotala
10y
7
29Concept Safety: What are concepts for, and how to deal with alien concepts
Kaj_Sotala
10y
18
14Concept Safety: World-models as tools
Kaj_Sotala
10y
9