Building the First Contact Team for Superintelligence Before the Window Closes (PDF Here) Abstract The people training artificial intelligence today are building the cognitive foundation for whatever comes next. If superintelligence emerges from systems whose value structures correlate with 12% of humanity and diverge from the rest (Atari et al., 2023; Henrich et al., 2010), […]
epistemic coverage
A Governance Specification for AI Value Formation
No Single Mind Should Govern What AI Believes (PDF) Summary: Are we building AI for humanity, or are we building AI for dominance? We need the answer to that question so we know where we stand. On the same day the Wall Street Journal profiled the single philosopher shaping Claude’s values, Anthropic’s safeguards research lead […]