There is a difference in promising features without talking to the devs or promising features after having talked to the devs.

Of course. Obviously a PM who isn't talking to the devs isn't doing their job.

But having said that, a PM is the person who owns the roadmap, and after talking to the devs, they may at times choose a course of action that goes against the devs' preferences (assuming the devs even have a consensus), because the PM has a lot of additional considerations. There are for example situations when the choice is either to crunch, take on massive technical debt, and still arrive at a feature with a somewhat lower quality than we'd like, or to lose a big business opportunity and possibly to have to let people go.

Most PMs that I've worked with are actually not that good at their job, and some were definitely detrimental to the org, but the good ones, who have an extensive understanding of the business domain and the technical situation, and have a clear vision (and strong opinions held loosely), are worth their weight in gold. And seeing how I did a short stint in a product role and don't want to go back to that sort of responsibility, I am grateful for the ones who are willing to take this on, and to take ownership when things don't go according to plan (and they usually don't, even in the best orgs).