What role will be predominant in 2030?

I don't see the scrum framework going away anytime soon, because it's simple, popular, and easy to combine with other practices. I'm sure it will continue to adapt, though.. Back in 2010, I could not have predicted where I'd be today, so there's no way I could predict what will happen in the next decade. Perhaps Scrum will evolve into something else entirely, or morph into something we don't recognize today.

XP is still popular in theory, but in my limited experience I have not witnessed many organizations bold enough to adopt it. I'm always curious to hear from anyone who's been in a committed XP shop, if you have stories to share.

I see AI taking some of the "guesswork" out of PM when it comes to budget, scope, schedule, and risk "estimates". Why are most large construction projects over budget and behind schedule? "We" always seem to be surprised when such events occur. Using AI, and harvesting data from similar type projects, much more realistic estimates could be made, because by nature, human beings always tend to be optimistic. There would be no such issue when using AI.

XP was as popular as Scrum in the software community in early 2000s. Many practices and concepts survived within Scrum. I agree that Scrum is simple and there is a huge certification industry that will support his survival but there are already signs of decline.
The main reason is that some teams pretend to do Scrum and when they fail the framework gets a bad reputation. Another reason for decline is that people have unrealistic expectations, like you can start without a vision, without understanding what you want to achieve. The Scrum and Agile are not imposing details but you still need a visionary (the PO) to succeed.

We implemented XP, almost by the book, in 2002. I was only an analyst/programmer at the time because as a new migrant I was advised that for managerial roles you have to wait. As an observer, with many years as a a dev manager, i believe that I can understand why that company went back to planned approach. The main reason, that will also impact Scrum, was lack of focus on efficiency. We did a lot of prototypes, sometimes without any use for the product, just because it was fun and interesting. We tried to build the tests (nunit) before the actual code but testing is an overhead, it is not reflected in the actual product and at some point the owner of the company asked us to focus on the functionality.
If you want to know more about XP I strongly recommend Ron Jeffries' site https://ronjeffries.com/. He is one of the XP authors, and a practitioner.

In principle I agree with you but... the reality is much harder. Perhaps this is the rot cause of (construction) projects going over budget and schedule. The scope changes and even if the scope doesn't change the environment will pose unforeseen challenges.
reality can't be modelled, at least in our lifetime. Look at the autocomplete AI in gmail. It 'learns' from us but it works only for very simple statements and most of the time instead of helping it'a waste of time.
I built a construction estimation software and I have pretty good experience with project estimation using Function Points, probably the most reliable technique.
In principle I agree with "harvesting data from similar type projects, much more realistic estimates could be made" but speaking of reality, I am looking for many years for reliable data on Agile projects. For traditional ones there are few good repositories but there I couldn't find anything for Agile. One reason is that there are no consistent metrics.
"human beings always tend to be optimistic.", yes, that's true, especially for 'developers'. Their estimation is always 60-70% less than the effort required to complete the project. That's where a skilled and experienced PM adds value. There are many (real) risks that can't be mentioned in the risk register because of their political sensitivity. One, from my experience. that tops the list, is the Sponsor's commitment to fail the project. For political or personal reasons some projects start without any support from the Sponsor (PO).
"AI won't pick this." AI will model the probable outcome based on an ideal world. It can't model (yet) human creativity, good or bad, which is an important factor in determining the success of a project.

Thank you. I agree that BA is a good option for Scrum Master, especially for (real) projects. Apart from the good facilitation skills, it has the best understanding of the product from the business point of view and can help the PO in managing the backlog.
That's one reason for recommending that PMs, where possible, combine the role with the SM within the Scrum team. Most PMs have BA experience, defining requirements, facilitating workshops etc.