4. From How to What

For most of computing's history we have given machines imperative instructions — step-by-step descriptions of how to perform a task. The alternative is declarative: you state the result you want and leave the system to work out how to reach it. Any engineer who has written SQL, a build rule, or a Kubernetes manifest already thinks declaratively in patches. What capable AI has done is push that line — between how and what — across almost everything we do. Increasingly you specify the outcome and judge what comes back, and the machine determines the method. Control migrates from how to what.