It's not that clear cut. Just last week Don Knuth, who is like the OG computer scientist, wrote a paper about how LLM basically solved the problem for him which he and his friends couldn't - https://cs.stanford.edu/~knuth/papers/claude-cycles.pdf. And Don Knuth was big AI sceptic before.
But he is very inteligent obviously so he could use AI efficiently. Writing forum software is basically, apart of the design of it, an integration problem - how to reuse already written pieces of software. This is the biggest problem today and people in AI world are trusting this is going to get solved, I believe.
