UX changes are more forgiving than API changes.

A few weeks ago I mentioned that when you change a UX you change the interface, meaning more users might notice/be confused (because the change will at least show up in people's peripheral vision if it's on a top-level screen, whereas a user of an API might not notice at all other parts of the API changing)

However, there's a countervailing factor:

UX is for humans, and humans are alive.

APIs are for code, and code is built.

Yes, a human wrote the code (probably) but once the code is written, it might not be changed for a long time even while it executes often.

If you were to sample from all code being executed across the universe at any given time, the vast majority of code would have had orders of more time of accumulated execution wall-clock time than the amount of wall-clock time of humans thinking about it when they wrote it in the first place.

This is part of the extraordinary leverage of code.

(There is code for which this is not true; e.g. little random glue code, but it wouldn't be prevalent in your sample because it's rarely executed)

Code can't adapt itself to integrate with an API; a human can adapt themselves for a changed UX interface.

There's an entity with agency at each time step of interaction; the human is always in the loop with UX, while with code the human is often only indirectly / rarely in the loop.