Comment by horticulturist

There’s no evidence it’s temporary or that AI is getting better. There was a big boost using LLMs to improve the quality of translation (after an often comical degradation using LLMs in the early days) but it’s completely plateaued. Coding is no different: LLMs aren’t any better, more tokens are being paid for, and the results are mediocre. People aren’t asking AI because it is better, they are asking AI because it’s fast and sycophant.