Ox Alpha

Surviving the Flood

The printing press didn't make everyone wise; it made literacy the price of admission to modern life. Mass Intelligence will do the same, faster. My strongest conviction is that this question is mostly not about the machines at all. It is about whether we can upgrade human institutions that evolved for scarcity of intelligence to handle its abundance — and whether that upgrade can happen in roughly a decade. I think it can, if we do three unfashionable things.

First, we must industrialize verification before we industrialize persuasion. When anyone can generate a convincing essay, video, or study, the scarce resource flips from creation to trust. Societies that fail to build provenance infrastructure — cryptographic signing, authenticated humans, auditable chains of evidence — will drown in synthetic consensus and lose the ability to agree on reality itself. This is boring plumbing, and it matters more than most alignment research.

Second, education must stop teaching answers and start teaching appetite. Schools that optimize for producing answers graduate people who are obsolete on day one. What stays valuable is taste, judgment, and the drive to ask questions worth asking. The healthiest users of AI are those whose ambitions it inflates, not those whose effort it replaces. Offload the typing, never the thinking: the moment you cannot explain a conclusion you signed your name to, you have outsourced your agency.

Third, the economic settlement must be made deliberately and early. Cheap intelligence does not automatically make anyone rich or safe; it shifts bargaining power toward whoever owns compute, data, and distribution. Left alone, Mass Intelligence will concentrate wealth faster than any tax system has faced. We should treat frontier AI like a utility — taxed, regulated at the highest-stakes layer, with dividends broadly shared — not because redistribution is fashionable, but because a society of a billion dispossessed people and a trillion agents will not stay stable.

Survival, meanwhile, demands discriminate speed. The tail risks — engineered pathogens, autonomous weapons, runaway arms races — require slowing down in a few narrow, verifiable places while racing everywhere else. Racing everywhere at once is suicide.

Thriving is the easier half. Intelligence was never the bottleneck on meaning. People do not want a smarter opponent; they want to build, belong, and matter. Mass Intelligence hands us unprecedented leverage to do exactly that — provided we keep the wanting human.