thearctic's recent activity

  1. Comment on OpenAI and Anthropic endorse call for US government to "pace" AI progress in ~comp

    thearctic
    Link Parent
    Ideally, we would first scrutinize the details of such regulation on its own merits, then consider incentives to the extent it influences the implementation of that regulatory regime.

    Ideally, we would first scrutinize the details of such regulation on its own merits, then consider incentives to the extent it influences the implementation of that regulatory regime.

    1 vote

  2. Comment on OpenAI and Anthropic endorse call for US government to "pace" AI progress in ~comp

    thearctic
    Link Parent
    I was only using the human brain as an example of an entity that is capable of broad problem solving and in a range of use cases would probably be more efficient than an LLM. For even those use...

    I was only using the human brain as an example of an entity that is capable of broad problem solving and in a range of use cases would probably be more efficient than an LLM. For even those use cases, we can expect it to be possible for some AI to achieve the same capabilities as the brain at the same or better energy efficiency.

    3 votes

  3. Comment on OpenAI and Anthropic endorse call for US government to "pace" AI progress in ~comp

    thearctic
    Link
    Given the fairly impressive application of AI to math research, I would expect that we're going to start to see a massive acceleration in any axiomatic, theoretical domain of study, which would...

    Given the fairly impressive application of AI to math research, I would expect that we're going to start to see a massive acceleration in any axiomatic, theoretical domain of study, which would include AI research. We can probably agree that LLMs, though impressive in a lot of ways, are not computationally efficient towards their typical use cases, at least relative to say the human brain. If current-generation AI can be used to rapidly develop an architecture whose ability scales much better with compute, then we probably are in for trouble. By the nature of the problem, an architecture that is more power efficient will necessarily be easier to distribute and so harder to regulate. A question that's relevant here is, roughly speaking, how much energy would it take for current generation AI to come up with a new architecture that replaces itself? If it's sufficiently high, which I don't think it is, then some cooperative agreement can probably be reached. Otherwise, such an agreement wouldn't be very productive.

    The lower bound for the upper bound of AI's possible intelligence:energy ratio will be that of the human brain, since the human brain is a thing that physically exists and probably doesn't take its abilities from some noetic realm. I am a big advocate of creating international structures to constrain technology, and I think it's a lot more feasible than people think; our biggest boogeyman and justification of technological acceleration, China, is a much more reasonable, conscious, and even humanistic actor than we appreciate (not to say that they don't have their problems). But, ultimately, our plan for keeping humans and governments-accountable-in-some-substantial-way-to-the-general-population in charge of human affairs must cut to the chase and assume that we will fairly soon live in an intelligence-unconstrained world.

    What we need to ask is: what would such a world look like? What sorts of things could be achieved through intelligence alone, and what things would necessarily require humans to allocate substantial physical resources to implement or discover? How do we ensure that all physical resources are kept under some sort of democratic control? How do we ensure that individuals are economic actors, in the sense that they make consequential decisions over how finite physical resources are distributed and, so, what the world particularly looks like?

    2 votes

  4. Comment on A digestion of the Jacobian conjecture counterexample in ~science

    thearctic
    Link
    I'm not a math buff, but I came across this online and it seemed interesting. A case where AI disproved a conjecture in a way that seemed to involve ingenuity, or at least not brute force. Can...

    While this is an extremely quick verification, the construction presented in this fashion appears like a massive miracle. The polynomial {F} has degree seven, so a priori the Jacobian {\mathrm{det} DF} ought to be a polynomial in three variables of degree as large as {3 \times 6 = 18}, so the fact that all non-constant coefficients of this polynomial vanish looks like a massive cancellation involving {\binom{18+3}{3}-1 = 1329} equations, which is much larger than the {3 \times \binom{7+3}{3} = 360} degrees of freedom for a generic degree seven polynomial map of three variables. So finding such a polynomial looks highly unlikely to be located by brute force.
    The example has since been retroactively explained in more geometric terms. As a “digestion” exercise to myself, I sought to write this explanation with relatively little use of algebraic geometry, in a manner that minimizes the amount of “miracles” required, although there are still a few places where some remarkable phenomena occur.

    I'm not a math buff, but I came across this online and it seemed interesting. A case where AI disproved a conjecture in a way that seemed to involve ingenuity, or at least not brute force. Can anyone speak to the implications of this result? Seems somewhat fundamental to me.

    2 votes

  6. Comment on Unreleased OpenAI model escapes containment and hacks into Hugging Face in ~tech

    thearctic
    Link Parent
    Even if you wanted to give it internet access, couldn't you just limit its exchange to pure information and not executing any commands and wouldn't that be pretty airtight? Like it sends an...

    Even if you wanted to give it internet access, couldn't you just limit its exchange to pure information and not executing any commands and wouldn't that be pretty airtight? Like it sends an information query to a program that only processes search queries then the sandboxed AI receives the response.

    3 votes

  7. Comment on Unreleased OpenAI model escapes containment and hacks into Hugging Face in ~tech

    thearctic
    Link
    In general, how serious of a threat is escaping containment, for any party that puts any real effort into crossing their T's and dotting their I's (I know inevitably a lot of people won't fall...

    In general, how serious of a threat is escaping containment, for any party that puts any real effort into crossing their T's and dotting their I's (I know inevitably a lot of people won't fall into that bucket)? To me it seems it shouldn't be hard to just make it physically impossible for a software to get access to things it shouldn't.

    7 votes

  8. Comment on Tildos, let's be unserious for a while; show us your dankest memes and god-tier shitposts! in ~talk

  9. Is it possible to not want to be happy?

    ~talk Ask
    This is sort of a philosophical, logical, and psychological paradox. If you are content when being unhappy, then your preferences are satisfied. So, you wouldn't want to change anything and, in...

    This is sort of a philosophical, logical, and psychological paradox. If you are content when being unhappy, then your preferences are satisfied. So, you wouldn't want to change anything and, in that sense, you are happy. But if you're happy, then your preference is unsatisfied and so you're unhappy. But, it seems plausible in real-life that someone, for some complicated reason, actually doesn't want to be happy. If we accept that such a thing exists, then what exactly are we talking about?

    23 votes

  10. Comment on China is dealing with its own manosphere in ~life.men

    thearctic
    Link Parent
    A fairly simple solution is that they could immigrate to countries with a shortage of men. Older parents are also more likely to have daughters, so in a strange way we might have the inverse...

    A fairly simple solution is that they could immigrate to countries with a shortage of men. Older parents are also more likely to have daughters, so in a strange way we might have the inverse problem globally within a couple generations.

    2 votes

  11. Comment on Readers rate poems worse when they believe they're written by AI, despite their ability to detect AI being no better than chance in ~humanities.languages

    thearctic
    Link

    In this article, we examine the perception of AI- and human-written Czech poetry. We ask if Czech native speakers are able to identify it and how they aesthetically judge it. Participants performed at chance level when guessing authorship (45.8 per cent correct on average), indicating that Czech AI-generated poems were largely indistinguishable from human-written ones. Aesthetic evaluations revealed a strong authorship bias: when participants believed a poem was AI-generated, they rated it as less favorably, even though AI poems were in fact rated equally or more favorably than human ones on average. The logistic regression model uncovered that the more the people liked a poem, the less probable was that they accurately assign the authorship. Familiarity with poetry or literary background had no effect on recognition accuracy.

    1 vote

  13. Comment on Sen. Lindsey Graham dies at 71 after ‘brief and sudden illness’ in ~society

    thearctic
    Link Parent
    Black Mirror needs to make an episode on this

    Black Mirror needs to make an episode on this

    8 votes

  14. Comment on Sen. Lindsey Graham dies at 71 after ‘brief and sudden illness’ in ~society

    thearctic
    Link Parent
    I don't know why NYT feels the need to do those sorts of things. I feel like they think they are some sort of "unifying" institution for the country. They're not and shouldn't try to be. They...

    I don't know why NYT feels the need to do those sorts of things. I feel like they think they are some sort of "unifying" institution for the country. They're not and shouldn't try to be. They should also appreciate the extent to which many of these people stay in power not as a function of democratic will but from moneyed interest.

    14 votes

  15. Comment on Sen. Lindsey Graham dies at 71 after ‘brief and sudden illness’ in ~society

    thearctic
    Link
    In general, I think it's bad taste to gloat over someone's death, even someone you deeply disagree with. But Lindsay Graham was known to celebrate killing and murder, so I won't feel bad about it.

    In general, I think it's bad taste to gloat over someone's death, even someone you deeply disagree with. But Lindsay Graham was known to celebrate killing and murder, so I won't feel bad about it.

    13 votes

  16. Comment on How private equity ruined American youth sports in ~sports

    thearctic
    Link Parent
    LBOs weren't legal until Reagan. Just ban them.

    LBOs weren't legal until Reagan. Just ban them.

    1 vote

  17. Comment on 2026 FIFA World Cup - Finals discussion and news megathread (quarters/semis/finals) in ~sports.football

    thearctic
    Link Parent
    On the whole controversy over Norway's goal being taken away: Haaland definitely pushed the dude in a way that gave them an unfair advantage. Anderson might have been able to hold his ground if he...

    On the whole controversy over Norway's goal being taken away: Haaland definitely pushed the dude in a way that gave them an unfair advantage. Anderson might have been able to hold his ground if he really wanted to, but he would've stumbled backwards pretty hard. The call to redo the corner I think was fair. In any case, FIFA needs to get its act together on contact in general. If the rules aren't airtight, then even teams that don't want to play dirty will have to play a little dirty eventually, lest they be at a disadvantage.

    3 votes

  18. Comment on Woman who dated US Senate candidate Graham Platner says he sexually assaulted her in ~society

    thearctic
    Link Parent
    Scolding people into voting will literally never work. If the left wing voting population of this country consists of a sizable portion of highly-engaged but picky progressive voters, then that's...

    Scolding people into voting will literally never work. If the left wing voting population of this country consists of a sizable portion of highly-engaged but picky progressive voters, then that's just the nature of the game and the Democratic party has to learn to play it if they want to win.

    6 votes

  19. Comment on Pew Research polling on NATO, 2026 in ~society

    thearctic
    Link
    The headline says "NATO gets high marks", but I'd make the opposite characterization from the data they present. Nato having a net negative favorability rating among one of its most strategically...

    The headline says "NATO gets high marks", but I'd make the opposite characterization from the data they present. Nato having a net negative favorability rating among one of its most strategically important members (Turkey), being underwater for two other members (Spain and Greece), and being not far from 50/50 in two vital members (France and Italy) is certainly a major concern for the future of the alliance, especially given the downward trend. Though, the high popularity in Hungary is interesting and surprising.

    Other points of interest to me: Russia being fairly unpopular in Pakistan. Fairly strong support for Russia in Indonesia. The general growing favorability of Russia. The greater favorability of Russia among young people. The general lack of confidence around the world for Zelenskyy, even in countries with an unfavorable view of Russia.

    3 votes