A convening of more than 45 AI experts concludes that openness can support AI safety, and provides a research roadmap, but offers no new empirical evidence.
Many real-world risks stem from misspecified user requests
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
-
A Different Approach to AI Safety: Proceedings from the Columbia Convening on Openness in Artificial Intelligence and AI Safety
A convening of more than 45 AI experts concludes that openness can support AI safety, and provides a research roadmap, but offers no new empirical evidence.