Login
You're viewing the front-end.social public feed.
  • Greg Wilsongvwilson@mastodon.social
    Jul 28, 2026, 11:28 AM

    In his essay on Salvador Dali, Orwell argued that because Dali was a repulsive human being, the right wouldn't admit that he was a great artist; conversely, because he was a great artist, the left wouldn't admit he was a repulsive human being. I'm seeing the same thing with AI: because it's unethical, one side won't acknowledge that it's useful, but because it's useful, the other side won't acknowledge that it's unethical.

    Time to make another cup of tea…

    💬 27🔄 171⭐ 288

Replies

  • 💬 1🔄 3⭐ 31
  • 💬 1🔄 2⭐ 22
  • 💬 2🔄 2⭐ 24
  • tinkeltinkel@mastodon.social
    Jul 28, 2026, 12:14 PM

    @morten_skaaning @gvwilson

    PROMPT: "Write few one-sentence sarcastic reactions to this post, written in the typical Mastodon style featuring strong moral framing, skepticism toward enlightened-centrist analogies, tech fatigue"

    Image attached toot
    💬 0🔄 0⭐ 13
  • 💬 0🔄 0⭐ 13
  • Gadfly (-booq-)gaditb@icosahedron.website
    Jul 28, 2026, 6:20 PM

    @morten_skaaning @gvwilson I can't tell if you're being sarcastic or earnest (or both?) but honestly I've been trying for at least a year or two to convince friends/acquaintences that antisemitism is used because it is useful, when explaining it at a deeper level to them.

    (Antisemitism isn't special here, I just explain it a lot.)

    So like, Yes.

    💬 0🔄 0⭐ 2
  • Lien Raglienrag@mastodon.tedomum.net
    Jul 28, 2026, 11:33 AM

    @gvwilson

    I've read a comic book that didn't name Dali but was clearly referencing him, and suggested that Dali used the Civil War in Spain (where he was protected by Franco's regime) to commit murders.
    Any idea if it's pure invention or if it's based on something true ?

    💬 0🔄 0⭐ 1
  • TrueNorthSpice 🇨🇦TrueNorthSpice@mastodon.world
    Jul 28, 2026, 11:58 AM

    @gvwilson

    It does have uses and the people I know, including myself, are angry it was released for public use with zero safeguards!!!!
    So please stop lumping people together and saying one side or another is having a knee-jerk reaction

    💬 0🔄 0⭐ 5
  • naturepokernaturepoker@genomic.social
    Jul 28, 2026, 12:49 PM

    @gvwilson I feel similarly. The 'no LLM at all costs' voices in the media are leaving a big wide gap that could have used positions calling for restraint and cutting down on hype.

    How the current generation of llms were developed and are being used is revolting, but the comments you've gotten here so far don't look a ton better.

    💬 0🔄 0⭐ 5
  • Adrianoadriano@lile.cl
    Jul 28, 2026, 3:20 PM

    @naturepoker @gvwilson
    "How the current generation of llms were developed and are being used is revolting, but the comments you've gotten here so far don't look a ton better."

    Please understand the depth, the big, wide gap, between those two concepts you just typed. Please notice the difference between comments mocking a single person on mastodon for having a hot take and "How the current generation of llms were developed and are being used is revolting".

    💬 1🔄 0⭐ 4
  • naturepokernaturepoker@genomic.social
    Jul 28, 2026, 3:25 PM

    @adriano @gvwilson ah yes. How stupid and short sighted of me. I'm terribly sorry to expose you, an insightful and superior being, to my lesser take. If only I had the benefit of your wisdom!

    💬 1🔄 0⭐ 3
  • 💬 0🔄 0⭐ 1
  • 💬 1🔄 0⭐ 2
  • Jason Orendorffjorendorff@federate.social
    Jul 28, 2026, 2:27 PM

    @Spirit @gvwilson It takes 53 kJ to heat 6 oz of water from 25°C to 100°C.

    There's no credible public information on how much energy an average ChatGPT query costs, but the high-end estimate is 3 Wh = 10.8 kJ. (Of course, OpenAI could just tell us, if they cared.)

    If 3 Wh is right, Spirit's claim exaggerates reality by a factor of 50.

    It might be more like 0.3 Wh, which would mean the claim is off by a factor of 500.

    💬 1🔄 0⭐ 4
  • 💬 1🔄 0⭐ 0
  • SpiritSpirit@cybernetic.monster
    Jul 28, 2026, 3:11 PM

    @jorendorff @gvwilson isn't 53 kilojoules ~.0147 kilowatt hours? So yeah I'm way off... But after plugging in the construction and operating costs I'm sure it's not going to look good.

    The big mistake is ignoring initial costs; how much energy did it take to get the GPU from silicon to inside the rack, etc.

    But we can argue about if it's one cup of tea or one hundred. Our electricity bill is up 16%. The planet is still burning.

    💬 0🔄 0⭐ 3
  • James WidmanJamesWidman@mastodon.social
    Jul 28, 2026, 1:13 PM

    @gvwilson it works so well that a typical LLM user apparently no longer feels the need to exercise the parts of their brain that were responsible for considering environmental costs, human costs, opportunity costs (as in, "how else could these trillions of dollars have been spent?"), etc.

    or maybe they were never inclined to care about that stuff in the first place.

    💬 1🔄 0⭐ 6
  • James WidmanJamesWidman@mastodon.social
    Jul 28, 2026, 1:19 PM

    @gvwilson like even if this shit could work perfectly without:
    - massive energy costs
    - theft
    - constantly re-scraping the web so hard that it kills the web
    - making PC hardware unaffordable
    - introducing any disturbance to people who now live next to a data center
    - etc

    ...even then, there would be the problem that people are using it as a substitute for all cognitive exercise, leading to a general atrophy and learned helplessness.

    💬 1🔄 4⭐ 14
  • James WidmanJamesWidman@mastodon.social
    Jul 28, 2026, 1:22 PM

    @gvwilson but also... what does "work perfectly" mean when the cost of *producing* a popular open-weight model is so high that only large companies would pay for it?

    seems like they're now in an excellent position to influence the personal politics of the LLM user.

    💬 1🔄 1⭐ 5
  • James WidmanJamesWidman@mastodon.social
    Jul 28, 2026, 1:28 PM

    @gvwilson i mean, they have a captive audience: they have a user who now cannot imagine living without their LLMs.

    and the LLM is so smart, and it's been getting smarter every day, right?

    so when it tells me that a particular ethnic group eats their neighbor's pets, or when it tells me about reasons why a new war was necessary, or why we can't afford to provide medical care to everyone, it must be telling the truth, right?

    💬 1🔄 2⭐ 6
  • 💬 0🔄 0⭐ 5
  • contrasocialcontrasocial@mastodon.social
    Jul 28, 2026, 1:47 PM

    @gvwilson

    AI is a marketing term. It's just a hype term for machine learning done at a hyper-destructive scale to funnel money to the ultrawealthy.

    There is no use that justifies that. None. It's not a bothsides or a failure to admit to usefulness. "AI" is not useful, it's counter to life on this planet.

    💬 0🔄 0⭐ 12
  • 💬 1🔄 0⭐ 0
  • LRlritter@mastodon.gamedev.place
    Jul 28, 2026, 2:11 PM

    @gvwilson i do think the observation depends on nutpicking extremists.

    namely, the subgroup of AI luddites in total denial of usefulness, and the subgroup of AI zealots in total denial of cost.

    if you were not in denial, then you would have to approach the truth from an economical angle: a cost-benefit computation.

    here you will also find a bipolar continuum: from short term benefit to long term benefit.

    💬 3🔄 0⭐ 3
  • LRlritter@mastodon.gamedev.place
    Jul 28, 2026, 2:13 PM

    @gvwilson going strictly by short term cost/benefit, the calculation is pro-AI: why do i need to bother that my codebase becomes unmaintainable or my services decay and backfire when i can be profitable today? but that is pretty much where zealots already are (prob unwittingly).

    the long term calculation is anti-AI: it's simply not worth the expenditure. and there you find common ground with the luddite again.

    so the passionate argument is just a shorthand for the economical one.

    💬 0🔄 2⭐ 3
  • 💬 0🔄 0⭐ 2
  • LRlritter@mastodon.gamedev.place
    Jul 28, 2026, 3:40 PM

    @s_d_brath @gvwilson they're both negative terms. y'all poopy poopies!

    no, seriously, i said "subgroup". we're talking of the end of the spectrum, where information doesn't matter and only personal feelings count. plenty of informed luddites with rational reasons (papers, studies, etc) and among the zealots i would guess the less radical ones are a little bit too optimistic as to the future viability of the thing, and overestimate the rapidness of the transformation we're experiencing.

    💬 1🔄 0⭐ 0
  • LRlritter@mastodon.gamedev.place
    Jul 28, 2026, 3:45 PM

    @s_d_brath at the other end of the spectrum there are luddites who use AI a little bit for stuff that doesn't matter, and zealots who use AI only a little bit for where it works, and those venn diagrams of usage probably overlap a lot but the fundamental stance as to the question "what would i do if this went away tomorrow?" would still be different.

    💬 1🔄 0⭐ 0
  • Braths_d_brath@mastodon.social
    Jul 28, 2026, 9:07 PM

    @lritter I am personally not opposed to the idea of AI. If all those data centers actually put up renewable energy matching their use, found a way to not deplete fresh water sources, and any financial gain mainly went to pay the creators that they stole from, I would be all for their existence.

    💬 0🔄 0⭐ 1
  • 💬 1🔄 0⭐ 2
  • LRlritter@mastodon.gamedev.place
    Jul 28, 2026, 6:45 PM

    @gaditb @gvwilson it's at least 10 years old. dozens of results for search.

    > Nutpicking is the act of selecting the most extreme or outrageous members of a group to represent the entire group, often to discredit or demonize them. This tactic is commonly used in political discourse to exaggerate differences and create division.

    💬 0🔄 1⭐ 5
  • 💬 0🔄 1⭐ 6
  • jorjor@foojay.social
    Jul 28, 2026, 2:35 PM

    @gvwilson It's useful if you switch off your moral compass, then it's superuseful...

    💬 0🔄 0⭐ 4
  • mancubemancube@pnw.zone
    Jul 28, 2026, 3:28 PM

    @gvwilson pretty sure "the left" knows dali sucks at this point and "the right" doesn't care. big misses in all directions on this one

    💬 0🔄 1⭐ 7
  • Damiano GacíkDamiano_Chech@techhub.social
    Jul 28, 2026, 3:43 PM

    @gvwilson Interesting analogy. I think we’re currently going through a stage where some people overestimate AI, while others underestimate it. As always, the truth is somewhere in the middle

    💬 0🔄 0⭐ 0
  • Phil Barkerphilbarker@mastodon.scot
    Jul 28, 2026, 3:43 PM

    @gvwilson You make it sound like Orwell sat on the wall, but he concluded that Dali's creations (or at least some of them) were "diseased and disgusting, and any investigation ought to start out from that fact"

    Same applies to LLMs?

    💬 0🔄 0⭐ 1
  • Anthony Soracea@pdx.social
    Jul 28, 2026, 4:55 PM

    @gvwilson @jminor I don’t think that’s true, overall. Certainly for myself, and many other people I have read, the point is we don’t *care* if it’s useful. I have a sound understanding of these systems current capabilities. But if the ethics behind them are crap, why does it matter?

    💬 0🔄 0⭐ 6
  • Isu 🐲Isurandil@mastodon.online
    Jul 28, 2026, 4:56 PM

    @gvwilson "AI" is not homogeneous. There are really good uses (e. g. all the ML stuff going on in the bio world) and really bad uses (ChatGPT & co. using the pre-AI slop internet without consent, fossil industry backed burning of the planet, ...).

    💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 2
  • Cindʎ Xiao 🍉cxiao@infosec.exchange
    Jul 28, 2026, 5:17 PM

    @gvwilson This is a view that assumes the only people in the world that exist are technologists, or at least biases "who matters" in the world heavily towards technologists.

    Whether us technologists like it or not, we live in a society. And that society will eventually come knocking at our door asking what useful things we have done with all the resources they have given us, and weigh it against what we have inflicted on them. They are already knocking, and the toll is already so unequal, as to cause the technologist's "reasonable middle position of usefulness" to appear as an incredibly naive view.

    💬 0🔄 0⭐ 4
  • Joshkrnlg@mastodon.social
    Jul 28, 2026, 5:20 PM

    @gvwilson
    That feels pretty real to me. I suppose the counter to this way of thinking is that it doesn't matter that it's useful because it isn't worth the harm. There are plenty of obvious examples of things like that, one I saw recently was Child Labour. People look in shock at how children worked in mills in the industrial revolution and got horribly maimed or killed. But it was profitable! *But totally wrong anyway*

    💬 1🔄 0⭐ 4
  • Joshkrnlg@mastodon.social
    Jul 28, 2026, 5:21 PM

    @gvwilson
    Yes it's useful. It's also destroying lives, the environment, eroding society, centralising power.

    So really the main use is what it gives the wealthy, not the little bit of convenience it can give to users.

    💬 0🔄 0⭐ 1
  • Lethelethe@digitalcourage.social
    Jul 28, 2026, 5:33 PM

    @gvwilson Congratulations, you have discovered the golden middle path towards enlightenment.

    In the meantime let me look at my books and old lectures on the scientific method and cognitive biases...

    💬 0🔄 0⭐ 1
  • [MARX_IN_SPACE]otherdog@mastodon.social
    Jul 28, 2026, 6:20 PM

    @gvwilson Despite the shit you’re getting for this I quite like these sorts of binary perspectives as thought experiments. Of course the world, the tech itself and people’s views are more complex and nuanced but this toot made a salient point that got me thinking and I appreciate it.

    💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 1
  • 💬 0🔄 0⭐ 0
  • Dave Wilburn :donor:DaveMWilburn@infosec.exchange
    Jul 28, 2026, 11:06 PM

    @gvwilson

    IMO its usefulness is still at least a little bit up in the air. There are too many legitimate concerns about user skill degradation, hallucinations, and business ROI at full unsubsidized token cost to confidently say it is useful. Perhaps its usefulness will be more firmly proven in the future as models improve, tool sets and processes around them improve, and the alleged usefulness is better studied. Or maybe not.

    In the meantime, while the uses are at best unproven, anecdotal, and speculative, the ethical problems and societal harms are much more firmly established.

    💬 0🔄 0⭐ 3
  • 💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 2
  • adritheonlyadritheonly@mastodon.social
    Jul 29, 2026, 4:24 PM

    @gvwilson Data centers are useful to make oil useful again now that private individuals are turning to solar power. Obviously "AI" is a very useful tool to make data centers absolutely essential.
    Even defending "AI" by any means is useful if you get paid to do so by those who stand to gain from an increased use of oil and more data centers.

    😒

    💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0