LLMs have really revitalized interest in these areas. AI can really help navigate the initial learning curve, can do a surprising amount of "heavy lifting" and can make tedious but useful work much easier. Do you want your little language to have a language server and nice editor-specific syntax highlighting? Do you need to write a parser with decent error messages? Do you need to prove a bunch of largely straightforward lemmas to get to the proof you actually care about? All of these things are easier (and, hopefully, more fun) than they were a few years ago. But, at the same time, there is still a lot of room for human insight and design in this process. There are a lot of areas that AI can't handle (or, at least, can't handle well) and, of course, nothing stops you from doing the fun stuff by hand even if you could hand it off to Claude.
And, of course, all this PL stuff was fun before LLMs. It's even more fun now even if you don't want to use AI yourself, because more people are doing and talking about PL stuff online, and there are more tools and libraries you can use yourself.