I feel sorry for the young hopeful data scientists who got into the field when doing data science was still interesting and 95% of their jobs hadn't turned over

I feel sorry for the young hopeful data scientists who got into the field when doing data science was still interesting and 95% of their jobs hadn't turned over into tuning the latest LLM to poorly accomplish some random task an executive thought up. I know a few of them and once they started riding the hype curve for real, the luster wore off and they're all absolutely miserable in their jobs and trying to look for exits. The fun stuff, the novel DL architectures, coming up with clever ways to balance datasets or label things...it's all just dried up. It's even worse than the last time I saw people sadly taking the stairs down the other end of the hype cycle when bioinformatics didn't explode into the bioeconomy that had been promised or when blockchain wasn't the revolution in corporate practices that CIOs everywhere had been sold on. We'll end up with this junk everywhere eventually, and it'll continue to commoditize, and that's why I'm very bearish on companies trying to make LLMs their sole business driver. AI is a feature, not the product.