Login
You're viewing the mstdn.social public feed.
  • Owlet of Minervanoctuaminervae@toot.si
    Jun 25, 2026, 1:48 PM

    The closest she gets to that positive vision is posing the question: How to tell humanity the story of "jobs changing for the better and new opportunities being created?" The question is left hanging.

    There's a whiff of vision over this line: "Sovereignty is crucial … because a Europe with agency can keep AI safe."

    Can it? In special economic zones so deregulated that US firms would flock there?

    In any case, that's still "avoiding the merely worse" of unsafe AI.

    3/n

    💬 1🔄 1⭐ 3

Replies

  • Owlet of Minervanoctuaminervae@toot.si
    Jun 25, 2026, 1:58 PM

    Dada writes, "Nobody adopts AI for the sake of adopting AI." I'm not so sure; it seems an economical explanation for many an AI adoption target set by CEOs or ministers.

    Still, I guess she's right. You can't convince voters by paraphrasing Douglas Adams' Prosser: "What do you mean, why's it got to be built? It's an AI data center. You've got to build AI data centers."

    But Prosser's argument is at least honest and spends no taxpayer euros on making up a Vision for bulldozing our home.

    4/4

    💬 1🔄 0⭐ 2
  • NoisecolorNoisecolor@toot.community
    Jun 25, 2026, 2:22 PM

    @noctuaminervae
    "we can't afford to fall behind" argument is weak. But it's not true that there are no upsides. AI has real potential to improve science, education and productivity. The real issue is whether that "potential" will be widely shared enough to justify the very visible costs of energy use, disruption, and job displacement. But then you do need strategic autonomy, but (again) in our car, whose autonomy french(?), since they are the only one with the electricity. Difficult situation.

    💬 2🔄 0⭐ 0
  • Owlet of Minervanoctuaminervae@toot.si
    Jun 26, 2026, 10:06 AM

    @Noisecolor I think my point was just that the column doesn't even articulate any real potential to improve anything. Let alone provide a justification in terms of the distribution of benefits.

    💬 0🔄 0⭐ 0
  • jaKa Močnikjkmcnk@mastodon.social
    Jun 26, 2026, 11:03 AM

    @Noisecolor @noctuaminervae to get any improvement, you first need to stop saying AI and start talking about statistical model based token string generation. only once you are understand how the hammer works, can you use it to facilitate some improvement with regard to nailing things. and so far, the smart autocorrect's improving capabilities have been confined to a few niche areas, hardly supporting the insane investment and environment degradation that the zealots want. :D

    💬 2🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0
  • NoisecolorNoisecolor@toot.community
    Jun 26, 2026, 12:03 PM

    @jkmcnk
    I just got from a long discussion about llms not being ai. But as far as I see it, to claim that you need to go against the established way we give meaning to words, then decide on a very specific and narrow definition of ai that is not shared by almost no one and against the people who have been working in the field for the last 50+ years. After all that, you would literally just succeed to call it something else for no real reason.
    @noctuaminervae

    💬 1🔄 0⭐ 0
  • jaKa Močnikjkmcnk@mastodon.social
    Jun 26, 2026, 12:14 PM

    @Noisecolor @noctuaminervae LLMs are NNs and therefore ofc part of all things AI. however, there's a wild gap between what a computer scientist calls AI (which has like half a century plus of tradition and includes expert systems, decision trees, neural networks and whatnot else) and what a random LLM peddler wants you to believe AI is (cf AGI delusions). thus, when you communicate with general public, one should be specific and avoid using the term AI.

    💬 1🔄 0⭐ 0
  • NoisecolorNoisecolor@toot.community
    Jun 26, 2026, 2:16 PM

    @jkmcnk
    Like I said, it's pointless, nobody cares and the only people who are interested in that are people who are vehemently against lmms, so there is 0 credibility.

    You become the "this is not real pizza" people. Or the "rock is not real music" or "games are not real art"...

    @noctuaminervae

    💬 1🔄 0⭐ 0
  • 💬 2🔄 0⭐ 0
  • 💬 1🔄 0⭐ 0
  • NoisecolorNoisecolor@toot.community
    Jun 26, 2026, 7:23 PM

    @jkmcnk
    I don't understand the need to do something catastrophic and I think it's unlikely. There are literally more than a trillion of dollars investments being planned in ai for next year.
    Even if it comes to a black swan biblical economic disaster, there will still be a need to run these models and there will be companies offering this service. Perhaps more expensive, perhaps more optimized models. And in two years chips will catch up and it will be like nothing happened.
    @noctuaminervae

    💬 1🔄 0⭐ 0
  • 💬 1🔄 0⭐ 0
  • 💬 1🔄 0⭐ 0
  • 💬 1🔄 0⭐ 0
  • NoisecolorNoisecolor@toot.community
    Jun 26, 2026, 9:52 PM

    @jkmcnk
    Which information?
    I just want to know how lmms are "taken care of".
    Cause I hear this a lot and there is never a description of how it happenes.
    I dont mean anything bad when Im curious about this.
    @noctuaminervae

    💬 0🔄 0⭐ 0
  • 💬 1🔄 0⭐ 0
  • 💬 1🔄 0⭐ 0
  • NoisecolorNoisecolor@toot.community
    Jun 26, 2026, 7:47 PM

    @jkmcnk
    I don't think the "LLMs aren't AI" framing will stick. It's not really true outside of a very specific technical definition, it doesn't reflect common language, and aside from a relatively small group of people who strongly dislike the technology, most people don't care about that distinction. It mostly seems to be a rhetorical argument rather than a meaningful one.
    @noctuaminervae

    💬 0🔄 0⭐ 0