sad serverWhy do you need petabytes of data and billions of GPUs for a model to occasionally produce the wrong answer?
Your executives are able to produce wrong answers much faster and more consistently
Kim@sadserver powered by a Happy meal
Andrew Douglass@sadserver scapegoats as a service
Mr. Completely@sadserver proving my point once again that these are the only jobs 100% ready for LLM replacement. Just train your model on Ayn Rand and YC. It probably only needs about a fifteen hundred word vocabulary as long as they're all pretentious but vague.
refraction 
@sadserver they are quite expensive to maintain though
VisualPlugin@sadserver: people produced computers in the 80s to occasionally break down and flip bits.
And then we iterated.
dynamicAnd all the executives need is three mansions, a half dozen sports cars, and a private jet.