Chronological feed of everything captured from Sebastian Mallaby.
AI governance faces an inescapable trilemma between national security, economic competitiveness, and societal safety, as pursuing any two objectives undermines the third. The singularity—a sudden intelligence explosion—is improbable due to human-controlled environments, physical infrastructure limits, and institutional frictions, necessitating gradual policy balancing over decades. Practical regulation should impose a "risk tax" on AI labs to fund safety research via tax credits, establish a revenue-generating national data repository under CAISI for model monitoring, and empower government veto over dangerous releases, fostering safety spillovers even amid U.S.-China rivalry.
Demis Hassabis, DeepMind co-founder, has dedicated his career to superintelligent AI aware of its world-ending potential, initially enforcing safety via Google acquisition terms like an independent oversight board and military bans. His singleton vision for unified safe development collapsed when rivals like Elon Musk founded OpenAI, sparking an uncontrolled AI arms race. Now a realist, he prioritizes gaining personal power within Google over formal governance to steer safety amid competitive pressures.