This is an example of a current trend: tools that help both humans and AI. Since both perform cognitive work, anything that makes that work easier benefits both.

TypeScript-first API framework built on the idea that an API's specification, implementation, validation, and tests should never drift apart, because they are the same thing.

Aren’t the generated tests superfluous since they just test whether the framework implements validation correctly? This should already be tested at the framework level and add no value?

Sort of, but the benefit is more along the lines of regression testing to make sure when things change they don't break. Also it does random fuzzing which can generate weird, edge-case inputs, like bizarre Unicode strings or extremely large numbers. These might technically pass declared schema validation but end up breaking your actual handler or database implementation.