Gregory's ServerThe field of #AI is such junk pseudo science at this point. Which other field has its equivalent of Nobel prize winners going absolutely bonkers? Between him and Hinton and Yoshua Bengio (his brother has the complete opposite view at least) clown town is getting crowded.
|
36 comments
@timnitGebru economics
0
0
20 Sep 2023 at 17:08 | Open on tines.spork.org
@timnitGebru I read that and I'm not a scientist, but it sounded like they calculated "being alive while creating art" as a CO2 creator? If I'm not wrong, that's a pretty scary proposal for how we reduce CO2 if AI is doing all our art and writing. @tjr @timnitGebru Yep. To compute the CO2 per written page, they took the CO2 a person emits per year by living and took the fraction representing the time required to write a page. As if humans would exist for less time if they did not write text. In case, anybody wants to see for themselves: https://arxiv.org/abs/2303.06219 @uncanny_static @tjr @timnitGebru There is a whole field of research (life cycle analysis) dedicated to assessing environmental impacts. None of the author seems to belong to that field. From what I read, this paper does not qualify as a proper lca. @uncanny_static @tjr @timnitGebru lol for the human writer they plug Mark Twain's output into the peak mean US carbon footprint where they should have plugged in a typical American writer penning SEO-filler clickbait for exposure in a country where 40% of the CO2e is emitted by the top 10%. @buermann @uncanny_static @timnitGebru "The quality might suck, but there's more of it! A win for AI!"  @uncanny_static @tjr @timnitGebru I only skimmed the “paper” but they don’t seem to be accounting for the time it took to develop the LLM software; the time to build, provision, maintain, and decommission its hardware; the time to develop the content comprising its training sets (though, fairly, one should divide this by the number of LLMs trained upon it…) All that to say, it’s a dumbass paper intended to reach a dumbass conclusion to be cited popularly by dumbasses.  Less carbon that someone drawing a smiley face in snow with their own pee? I’m skeptical. Presumably they’re charging the human creators for their own carbon footprints but the AI generators are being considered in a vacuum without charging them with the carbon footprints of all the humans who worked to develop the software and train them and build the hardware and maintain the hardware etc etc etc. @jonhendry @timnitGebru I do want to read what the paper say but I'm very skeptical about how the whole training and making the data set emit less CO2 than just... you know drawing and writing.  @timnitGebru wait, it’s literally the same misdirection cryptocurrency folks tried to make it look like banks were worse for the environment than blockchains? it really is all the exact same grift  @timnitGebru @zzt also what is this even measuring? “Writing one page” !? Is it a measure of typing speed? If i wanted to match the gpt results with text of similar depth and thought I’d just copy and paste from random sources and i bet i could “write one page” with the same “energy efficiency” … this is bs on so many levels. @zzt @timnitGebru this really is bonkers level shit. To reduce text and images to a dichotomy of existing or not existing is scary level talk from anyone, without having to go into sentimental discussion of what “creativity” means @zzt @timnitGebru and the energy debt in all those generative images and texts that need to be corrected @fasterandworse @timnitGebru on top of that, it’s extra bonkers and insanely dehumanizing that the fiat currency and financial systems crypto grifters used to rant about have been replaced by just ordinary human writers and artists in this iteration of the scam and of course, we’ve already seen the level of irrational hatred these folks had for fiat currency and banks @zzt@mas.to @timnitGebru@dair-community.social It would be fun to know how many former cryptomining GPU rigs have been directly repurposed to train and run LLMs and image generators. I bet it's a huge amount! Same grift, right down to the literal hardware supporting it.
@timnitGebru I think it's called Nobel disease https://en.m.wikipedia.org/wiki/Nobel_disease  @timnitGebru No other serious academic field today has a mainstream view that is bogus, pseudoscientific, or otherwise totally misguided, with serious scholars trying to subvert it from the margins, that much can be said about 'AI studies' @_siddharthsoni i highly disagree with this. if you think most modern sciences are devoid of bs then explain how we have a reproducibility problem or why much of @timnitGebru's findings on race and AI exist in the first place. institutional sciences have always existed to uphold the hegemony the institution is run by. until societies solve their historical use of science to marginalize/oppress their out groups, the sciences will always be flawed and inherently biased towards prejudice over fact. @wallamidyl @timnitGebru I agree that all mainstream science is hegemonic and tends to uphold colonial views until they are subverted, but pseudoscience of the kind peddled by mainstream AI studies is more than just a matter of upholding hegemony - it is simply and uncommonly unserious @timnitGebru @timnitGebru That kind of makes me want to read the paper just to see how hard they cooked the books to get the desired outcome.  @timnitGebru so most of their claims should be taken with a huge pinch of salt?  @timnitGebru Hold up... how exactly is "using generative AI" a distinct case from "... with the aid of a computer"? Maybe I should measure the carbon footprint of me doing basic proofreading for these clowns' bullshit... @timnitGebru Luc Montagnier, unfortunately. He can be directly credited with 15-20 000 COVID deaths from COVID alone in France alone. Thanks for passing the information - I knew for LeCun, but for others it was less obvious. @timnitGebru Jusult off the top of my head, William Schockley and Linus Pauling were Nobel Prize winning crackpots. @barryparr @timnitGebru another picturesque Nobel laureate @barryparr @timnitGebru Don't forget Brian Josephson as a more innocent example! Unfortunately the only Nobel Laureate whose lecture I've attended.  @timnitGebru |