@metalisp Why?
It seems to me, on the contrary, that #Lisp is in as good health as I have seen it any time these past fifty years. #Clojure is booming, and sprouting offshoots like #Scittle and Jank; Emacs Lisp and Scheme are still widely used; and there are lots of people playing with new, experimental Lisps. Common Lisp, even, seems particularly favoured by arts programmers.
It all looks pretty vibrant.
@simon_brooke @metalisp In a world where most code is throwaway and you don’t care about humans understanding it, language might not matter much.
In the places where code matters and communicating precisely with humans for understanding is critical, though, the language matters very much. And the Lisps have always shone there, and may be part of the current renaissance.
@benjamineskola @simon_brooke @metalisp
My theory is the more code is available to train LLMs in a particular language, the less "good" it will be at producing output that looks plausible, so fewer people will ask code generation tools to produce their desired ideas in that language.
AI boosters will say the language is "dying", but human skill in that language will be thriving more than the ones with steadily-increasing slop pollution.
@petealexharris @simon_brooke @metalisp I should add that part of my assumption here is that 90% of claims about "AI" are bollocks, and that's when I'm feeling generous.
But even if it's true that "AI" will eventually replace all programming, that clearly hasn't happened yet, and I haven't even seen signs of it starting.
(But maybe that's also because I'm not looking at the worst-affected languages. I think it probably also is true that if this *was* happening, Lisp would be affected less, or more slowly, than some of the more commercially-popular languages.)