Longer looks at where AI is actually headed, written from the floor of the industry rather than the bleachers.
Machines of Loving Grace, Two Years On
In 2024, Dario Amodei published Machines of Loving Grace, an essay about what powerful AI could do for the world if things go well. Reading it in 2026 means reading a prediction document while some of its bets are already being scored. This series grades it section by section, from 25 years of building conversational and agentic AI systems in production.
· The essay's engine is a country of geniuses in a datacenter compressing a century of science into a decade. Brute force can search any haystack, but only after someone defines the needle.
· The economics section asks the developing world to generalize from five confounded examples in a game that optimizes back. Development is a dataset problem, and the dataset is adversarial.
· The governance section, graded against the Silo novels. A coach is an intelligence you can ignore. A governor is the same intelligence with a pipe behind its recommendations.
· The final section asks the craftsman's question and misses the provider's. Meaning was never stored in the tasks; it is stored in the role the tasks fund.
The series is complete. I'll keep grading as reality keeps arriving.