boa00

Not sure I agree here

Text is just human thoughts in their most simple form. Writing is about expressing ideas, and there is almost an infinite number of ways to express them. Extremely difficult task, and LLMs only "imitate" it to the best of their training

This is not at all true for voice. There are an infinite number of possible voices, but a finite number of tones and phonemes you can use to express the text.

It's a much easier technical problem; it's just that it's much harder to gather proper data (you cannot just scrape Reddit and hope for the best, as LLMs do). And voice gets like 1/100th of LLMs' funding

Right now, yes. I have a job here, and I would get fired if start working from other country, so it's important to find a job first and then move

I agree with you. The feeling that grass is greener might play a role here, as I've already seen the 'ugly' side of DS, but not of SWE

But I thought I could easily come back to ML if I want to after SWE jobs given my prev experience. From inside the industry, I see that companies really want Data Scientists who know how to code properly

I can relate to this. The process of studying ML algorithms, how and why they work from the mathematical standpoint was really fun. But the process of building models themselves seems quite boring to me