When hn first started taking llms seriously maybe, 2-3-4 years ago (?) one of the most interesting recurring comments was that the real research goals in the fi

When hn first started taking llms seriously maybe, 2-3-4 years ago (?) one of the most interesting recurring comments was that the real research goals in the field would be reducing cost, doing more with fewer parameters, making lean models, and generally improving efficiency. Of course, those research goals run counter to the need to raise capital and do massive real estate land grabs for data centers, so they are not being prioritized by the top firms running the bubble game.