An empirical approach to AI safety #1 Post by tMcGrath » Thu, Jan 18, 2024, 1:19 PM UTC An empirical approach to AI safetybanburismus.substack.com