This article brings a very interesting angle about programming. It's fun to program in Python where code can be created with the speed of thought. This certainly broke some fun previous generation of programmers had with C or Assembly.
It's just fun goes next level at least at this moment. There might be some paradigm changes in programming in 2-3 years time, for instance, having AI-friendly programming languages or some kinds of auxiliary AI tools, which can solve problems in some domains.
But is repeating something already known always fun? In studying something - yes. Repetition can be good. But imaging in 4-5 years time programmers (and maybe also mathematicians?) will always need to be on the bleeding edge to be relevant! Because everything which is already known, will be in the memory of SuperAI, so plagiarism will be immediately obvious. And there might be new super-libraries, which will have all the solutions invented so far. That is actually an ideal of software industry.
That is, AI can bring software developers to be inventors all the time. Maybe, bosses will judge productivity not by lines of code produced (or whatever the code-related metric can be) but by genuinely new solutions. (I only hope this will not go down the drain by patent silos - because perhaps AI can also be judging novelty of solutions.)
That will be fun, not writing some utility scripts.
AI can also help realize ideas because sometimes some core idea require dull implementation details around it. More fun can be had by producing not only code and test for it but formal proof the code works.
And think about completely new "coder for hire" market where nerds will solve new problems AI can't do. (Think leetcode, but without ready answers). That is, there will be a stream of purely novel and interesting problems, not something already solved hundreds of times.
Philosophically speaking, now the art of programming will search for both new forms and new content. More things will be possible. It's a great opportunity also for having fun.
