Balaji Srinivasan: How AI Will Change Politics, War, and Money

Balaji Srinivasan argues that AI is better understood as a set of competing systems than as one inevitable superintelligence. He contrasts a “monotheistic” AGI story with a “polytheistic” one in which different cultures, countries, and communities each end up with their own models, norms, social networks, and even cryptocurrencies. In that framing, AI becomes an oracle inside a broader “network state” stack alongside deterministic crypto and social coordination layers.

He also pushes back on the idea that AI will simply reason forever and outmaneuver humans. In his view, real systems face hard limits from chaos, turbulence, and cryptography, and AI cannot reliably predict everything or act fully autonomously because it still lacks embodiment, self-prompting, and a closed control loop. The practical bottleneck is not just generation, but verification: prompting may be easy to scale, while checking outputs and grounding them in reality remains expensive.

The conversation repeatedly returns to where AI is strong today: language, visual generation, and user-facing workflows where outputs are easy to inspect. Balaji argues that AI is weaker in adversarial, time-varying domains like markets and politics, and that the near-term business opportunity is proctoring, verification, and human-AI amplification rather than full agentic replacement. He also makes the case that drone warfare, not chatbots, is the clearest form of “killer AI” already in play.