Yes, true advice for now, however that's in niches; there is not enough work there for millions, let alone 10s or 100s of millions coders. And the experience is also often painted either more positive or negative because of their experience; I don't know what OP specifically does, but I have been integrating LLMs into everything I have made the past 6 months; the results are amazing and I guess also scary for some. Claude Haiku has been solving problems for us that were absolutely in the category of 'not possible' a few months ago.

I wrote code for 44 years now, but this is the last year; even when you need to explain the hard parts to llm; I simply don't need to bother with code anymore. I get more robust and better results from what we have created. My fear is just IF llama or a better open model will get there soon enough before these companies close the taps. Llama works well, but it's just slower (it requires more agent generations than claude or gpt).

Edit: ah quants; yeah niche and hard. Not a very big market for many coders.