“It saves less than even the ‘small stuff’ that we can do, like recycling, reusing plastic bags and replacing our lightbulbs. These are worth doing, by the way, but not at the expense of the big stuff like diet, cars, home heating, and flights, which can often save tonnes of CO2 a year.”
Comments
fwiw, my view of of genAI ranges from meh to wtaf. But I'm dismayed by the insanity it seems to induce in otherwise thoughtful people.
https://scholar.harvard.edu/files/jenniferlerner/files/tetlock_2000_jpsp_paper.pdf
https://psycnet.apa.org/record/1997-04623-003
https://pubmed.ncbi.nlm.nih.gov/12860191/
I'd be concerned about an argument form that ruled out (or in) certain conclusions, though I appreciate that's wading into deeper, though v interesting waters.
(I think she wrote the new post because people keep asking "but what about the cost of a query"; note that the CO2 estimates have training included)
Quite bad analysis imo
Aggregate AI energy consumption is a big problem, but each individual query is just a drop in the bucket.
1) Just some guy. I'm trying to pull an argument together from pre-existing info and not claiming to be a primary source
2) Definitely not "pro AI." I'm mostly pretty concerned about AI, but part of that concern comes from thinking AI systems are currently pretty capable.