Comment by hackermatic
1 year ago
People and governments really need to start asking if the results of the current generation of FAANG-size LLMs are worth the costs, in additional chips, power, and cooling water -- especially in locations that are constrained for any of those resources. Are we getting ten times better Google results for ten times more energy? I don't think so.
That question is for the investors to worry about, as long as the public and government aren't the ones footing the bill for compute.