Skip to content

Writing & Essays

I write on LessWrong as azsantosk — I've been there since 2018, with a dozen posts and a few hundred karma.

The piece closest to what I do now is "The Market Singularity", which argues for securing human influence over AI through market and decentralized mechanisms rather than centralized control.1 In "Does AI governance need a Federalist Papers debate?" I make the case for adversarial institution design — governance with real power-limiting mechanisms.2 My most developed alignment piece, "An AI-in-a-box success model", argues that a safely boxed Oracle AGI is both easier to build and time-competitive with the unsafe kind.3

I've also written "Cruxes in Katja Grace's Counterarguments", decomposing the case for AI existential risk,4 and "Optimization happens inside the mind, not in the world", on why model-based agents optimize the map rather than the territory.5

All posts

  1. LessWrong — The Market Singularity A New Perspective ↩

  2. LessWrong — Does Ai Governance Needs A Federalist Papers Debate ↩

  3. LessWrong — An Ai In A Box Success Model ↩

  4. LessWrong — Cruxes In Katja Grace S Counterarguments ↩

  5. LessWrong — Optimization Happens Inside The Mind Not In The World ↩