By “choose not to be helpful”, do you mean that they don’t use AI to detect spam notifications and distinguish them from useful notifications? That seems like quite a difficult engineering problem, though I’m sure their new AI could help.

There are many tools at their disposal, of which AI may be only a small one.

1) They could implement a “promotional” notification category the same way that they do “time sensitive” and let the user customize that the same way, or at least offer an opt-out.

2) Add a “report” button next to the disable notifications button.

3) Stop delivering notifications on behalf of apps that don’t correctly categorize promotional notifications.

4) Skip all this complexity and just ban developers that send spam. This is what their monopoly on app distribution is supposedly justified by: protecting users from malicious app developers.