This only proves the original point which is that there is not much demand for actual machine learning expertise because that is only carried out in a small number of places and what demands there is is for the more basic software carpentry like infrastructure and operations rather than the actual technology and Engineering side of things

▲

embedding-shape an hour ago | parent [-]

What parent says about "there are very few available positions" for "engineers that can do real LLM machine-learning" is fair, yeah, I'd agree with this.

I don't think the "incredibly small minority of companies in the world do any real training or optimisation" part is necessarily as true, as some parts of the work I do get is about helping them optimize training and infrastructure around training. Mind you, none of this is for building LLMs from scratch, it's 99% fine-tuning existing checkpoints.

I'd also agree with "paulg is in somewhat of a bubble" regardless of this, which is worth remembering whenever you read his content. Same goes for any person living in SF, and dare I say the US. But also, YMMV, I live and work in Europe, probably why I have this perspective.

▲

mike_hearn 38 minutes ago | parent [-]

I think Paul lives in Britain these days.

	▲	TheOtherHobbes 20 minutes ago \| parent [-]
		I think he does, but it seems he still considers himself a remote SV native rather than someone working in the UK's economy. Worst of both worlds - no casual gossip feed in the Bay Area, no big dog meetings in the UK. (Which mostly has no idea he exists.) As for the question - what are the odds LLMs will be anywhere near the top of the tech tree five years from now? The trend seems pretty clear to me - local/offshore models are snapping at the heels of the big names in the US, and the current investment arc is insane. I wouldn't bet on Anthropic or OpenAI being leaders five years from now. Longer term, I especially wouldn't bet on the US build-yourself-a-monopoly corporate model surviving AI at all.