Question for AI/ML scientists out there, regarding the enormous energy needs for the tech. This should have been easy to calculate well before the technology started to get into production, so I'm surprised it's only starting to be publically considered a problem. I'm sure there has been research on computational complexity of ML algorithms, LLMs etc. Or is it more complicated than that ? Are we in a grey zone where we don't know how things are supposed to scale until we try ? @cazencott
@jaztrophysicist @cazencott I guess the main reason is that nobody gives a shit.
@jknodlseder @jaztrophysicist @cazencott Bingo. If we can spend the energy equivalent of a small nation on printing monopoly money, then why does it matter how much we spend on something actually useful?
