[jtmoulia]

Sigh classic LLM -- without you, the expert, I can't quickly tell from the code / output how the answer the LLM produced is wrong. I also asked it to solve sudoku by "propagating constraints" and the answer seemed to work for me :/ Again, I'd guess the soln produced is wrong because I trust you more than the LLM but I don't have the mental horsepower to figure it out without resorting to tests & debugging.

[climatologist]

If you are happy with the tool then continue using it. My point was a simple one, any generally intelligent software system would have given you an optimal solution by accounting for what people know about backtracking and constraint propagation but the solution presented did not account for how to optimally use coroutines to propagate constraints.

I'll repeat what I said previously in a different way, LLMs are useful but they are nowhere near what is required to achieve generally intelligent software systems. I'm sure they will continue to improve as engineers and companies learn how to utilize them in their workflows but let's temper the hype a little bit because statistical autocompletion is not enough to achieve general intelligence.