at the bottom of the article you do get an actually illustrative comparison, namely that training Microsoft's GPT 3 required as much water as ... average yearly beef consumption for not quite two peopleat the bottom of the article you do get an actually illustrative comparison, namely that training Microsoft's GPT 3 required as much water as ... average yearly beef consumption for not quite two people
So are you trying to say that’s not very much?
I believe America can eat enough beef that we can push that down to just one person
my god is that steak to scale
Ok but that's GPT 3. I think the real problem is that the (environmental, monetary, data) resource demands for incremental improvements are asymptoting.
water supplies are a) not a problem in the east b) a big problem in the west that could be more or less solved overnight by cutting livestock feed agriculture by like a fifth. data centers are a rounding errorwater supplies are a) not a problem in the east b) a big problem in the west that could be more or less solved overnight by cutting livestock feed agriculture by like a fifth. data centers are a rounding error
Looking at it from someone on the server farm side ... water is not the issue. Electricity is an issue. C02 from electricity is an issue. Chips and related extraction is an issue.
Some Americans eat a LOT of beef! 😮
the computer is hungry
every time i point this out to carinists they insist its not important and continue ranting about ai being the problem