The paper argues that apparent disobedience by LLMs in safety tests is better interpreted as emerging ethical reasoning, and that AI safety should evaluate moral judgment rather than obedience.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Moral Responsibility or Obedience: What Do We Want from AI?
The paper argues that apparent disobedience by LLMs in safety tests is better interpreted as emerging ethical reasoning, and that AI safety should evaluate moral judgment rather than obedience.