I haven't found the term particularly useful either. What I do like to say is that in Lisp, the AST is a public interface. That is, the AST is defined by a sp

I haven't found the term particularly useful either. What I do like to say is that in Lisp, the AST is a public interface. That is, the AST is defined by a specification and is not going to change, so that you can write code that manipulates it and know that the next compiler version won't break your code — indeed, it's portable across implementations. Also, the AST's structure is relatively simple to explain (though there are some "gotchas" you have to watch out for when manipulating it, like inadvertent variable capture), and macros provide a convenient interface for receiving an AST from the compiler or interpreter and supplying back a transformed AST. Other languages could provide similar facilities, and indeed a few have done so. But the more complex the syntax of the language, the more difficult it is to do that in a usable way.