A new benchmark elicits AI models' value priorities from choices in 3,000 AI-risk dilemmas and reports correlations between those priorities and risky behaviors, including on the external HarmBench.
Claude’s Constitution
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
other 1
citation-polarity summary
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1roles
other 1polarities
support 1representative citing papers
citing papers explorer
-
Will AI Tell Lies to Save Sick Children? Litmus-Testing AI Values Prioritization with AIRiskDilemmas
A new benchmark elicits AI models' value priorities from choices in 3,000 AI-risk dilemmas and reports correlations between those priorities and risky behaviors, including on the external HarmBench.