The New Laws of Intelligence: A Manifesto for the AGI Age

Source: Nate’s Newsletter Date: 2024-12-22 URL: https://natesnewsletter.substack.com/p/the-new-laws-of-intelligence-a-manifesto

Summary

Intelligence now scales exponentially rather than linearly — o3’s discontinuous capability jumps (breaking ARC-AGI, ranking #175 programmer globally, scoring 25% on FrontierMath) are evidence of a structural shift, not incremental improvement. The real constraint is organizational readiness, not technical capability: abundant AI meets scarce human ability to deploy it effectively.

Implications

AI economics thread. Dual and potentially triple scaling curves (training compute, test-time compute, distilled inference) suggest compounding intelligence growth — but the cost-complexity paradox is real: 3.5 percentage points of ARC-AGI score gains per hour at extreme compute thresholds means marginal intelligence improvements become exponentially expensive. Cost floors don’t vanish just because capability ceilings rise.

Vendor positioning thread. OpenAI’s o3/o3-mini tiered strategy targets specialized users — the capability is there, but the product decision is who gets it and at what price. This is a deliberate positioning choice, not a technical limitation.

Agent product strategy thread. “Undiscovered capability” and integration bottlenecks across industries are Nate’s framing for where the actual work is — not building smarter models but deploying existing capability into organizational contexts that aren’t ready for it.

Watch: Whether the organizational readiness constraint proves more durable than the technical capability constraint, and whether it becomes the dominant explanation for slow enterprise AI ROI.