You are technically correct and yet you are missing the original point that people expect the super-intelligent AGI of science fiction when they hear the term, no matter how much all those lesser forms are AI too by the definition of the scientific field.
Neither the position to keep all the old solutions because they are old nor to adopt all the new solutions because they are new is sensible.
Some old solutions worked in the past and don’t work anymore because the actual world around us changed (the bits outside our control, e.g. some resources might be more sparse but were more plentiful in the past, human populations are larger, the world is more interconnected,…).
Some old solutions appeared to work in the past because we didn’t have the knowledge about their flaws yet but now that we do we need new ones.
Some new solutions are genuine improvements, others are merely sold by marketing and hype.
Some new solutions have studies, data or even logic and math backing them up while others are adopted on a whim or even contrary to evidence or logic.
We can not escape the fact that the world is complex and requires evaluation on a case by case basis and simplistic positions like “keep everything old” or “replace everything old” do not work.
To be fair ActivityPub is a pretty shitty protocol in terms of scaling up with all the quadratic communication and caching growth it requires. Not that ATprotocol is better, just that there is room for improvement on ActivityPub before it could be used on a world-wide scale for the entire human population the way major social media sites are right now.
What kind of person makes their car brand their social media display name?
Keep in mind that the people advocating for AI in all kinds of fields without good reason now are the very same people who never liked the privacy rules in the first place.