Nick

Kudos to the AP here. One of the reasons I can't take much of the AI discourse (especially LLM boosters) seriously is because of how much of it is couched in anthropomorphic terms. For any questions related to AI and philosophy of mind, this is essentially begging the question and, given that humans have well-known biases in this regard, actively encouraging irrational thought. https://futurism.com/artificial-intelligence/ai-bros-meltdown-feelings

@jwildeboer@social.wildeboer.net I would welcome correction by those more knowledgeable, but for the benefit of people like me who were wondering, it looks like this just covers MPEG-4 part 2, i.e. the video codec that is roughly equivalent to H.263 and was used by things like DivX, not the more prevalent H.264 (which I believe is MPEG-4 part 10).

@autiomaa@mementomori.social So the bots have an option to bypass the captchas meant to catch bots but the humans don't. That tracks. 😩 @mark@mastodon.fixermark.com @david_chisnall@infosec.exchange

Recently, while helping an older person figure out how to do something on their smartphone I was struck by how utterly opaque they are. Older computers had many issues, but at least you generally had windows that told you which application you were looking at, menus and buttons with words on them so you could see the options you had and have some rough idea of what they would do, and often even a built-in help feature. On a smartphone it can be non-obvious what application you're looking at. UI is often accomplished by gestures that may have subtle differences and be difficult to describe or even replicate by observing. The small screen and precise touch required make it especially challenging for anyone with any limitations on sight or motor skills. Worse yet, other computer systems seem to be taking their cues from mobile OSs more and more (and even before that they had begun to move to opaque icons rather than words on their buttons). It's really unfortunate that smartphones have become the predominate form of computer. I had hoped that AR headsets might lead to much more usable UI for small portable computers (like smartphones), but I think Google and Meta have pretty well soiled the reputation of AR glasses.

@autiomaa@mementomori.social @mark@mastodon.fixermark.com @david_chisnall@infosec.exchange Honestly I'm kind or surprised there isn't a "pay Cloudflare for X connections without a challenge/captcha", because it would be another revenue stream for them.

@johncarlosbaez It seems very odd for a atheistic biologist to assert that consciousness is "for" anything. Or perhaps he means ..."what the hell is *the word* consciousness for?"

@vicgrinberg@mastodon.social @johnnythan@tuebingen.network This XKCD is more accurate than one might think. I've been guilty of it too.

@Thomaskirchner7@rollenspiel.social @Natasha_Jay@tech.lgbt

@darkuncle@infosec.exchange I need to read both, so I won't offer an opinion now, but I wonder how this squares with this result: https://www.nature.com/articles/d41586-026-02167-3

@Pineywoozle@masto.ai If it helps, some of them do pulsate (though I don't think this could ever be evident by eye). https://en.wikipedia.org/wiki/Cepheid_variable @vicgrinberg@mastodon.social @quixoticgeek@social.v.st

@SteveBellovin@infosec.exchange If only it were cricket rather than soccer perhaps we could have the 100 years game. 😜

@argv_minus_one@mastodon.sdf.org In fairness, for people who only have any memory of the 21st century I can understand how the idea of society coming together at scale and spending resources to tackle a foreseeable problem before it becomes a crisis might seem farfetched. @odr_k4tana@infosec.exchange @targetdrone@mastodon.social @sophieschmieg@infosec.exchange

@brouhaha@mastodon.social @nyrath@spacey.space The latter is usually the case for things headed to space, especially if you can't practically switch over to a redundant spare. Since you went to all the money and effort to put it in space, you want to make sure it works for the intended mission life with high confidence, so you add a fair bit of margin, which means it has a good chance to last quite a bit longer than the intended mission life. But, as @Ann_Effes@berlin.social points out, there are costs to doing that, so you try not to go too overboard. @pomarede@mastodon.social

@darkuncle@infosec.exchange Firstly, I believe that even in the current situation malware can still circumvent the protection, because that's how jailbreaking generally works (it's not malicious but the same mechanism could be used by something malicious). Granted I think jailbreaking mechanisms have become scarce more recently. I also think it's quite possible to arrange it so that the way to leave to the locked down mode is only exposed in a place that is out of reach by malware (except in the case some some serious exploit as for jailbreaking), e.g. during the boot process while executing signed code. I do think that there is a related issue that can't be avoided though, and that's scammers talking people into leaving the locked down mode under false pretenses. I don't believe it's possible to fully avoid that risk, though if you make the process complicated and time consuming that helps. @tbortels@infosec.exchange

@darkuncle@infosec.exchange I think it's fair to default to a very locked down setup and even to make it a bit onerous to change that, but going beyond that is, as @tbortels@infosec.exchange has said, asserting that you the manufacturer are the true owner of the device and the user is just borrowing it. I also think that the behavior is far more driven by commercial considerations than user security. https://appleinsider.com/articles/26/04/17/app-store-scams-are-getting-worse-and-apple-isnt-doing-enough With mobile devices becoming such a large proportion of all personal computing devices, the implications of this for personal freedom (from the whims of corporations and, by extension, from the whims of governments) are dire.

@tbortels@infosec.exchange @darkuncle@infosec.exchange I'm shocked that you guys don't trust Sony. 😉 https://indieweb.social/@chrisoliver/116822384708101756

@zwol Same, except I have actually tried to stop (just because it is unnecessary when not in physical print), but it's really hard; it's just deeply ingrained in my typing reflexes. @luca @draNgNon @brad @jonny

@grimalkina I think software engineers underrate how often they are actually talking about psychology (or human behavioral science more generally). In my experience, it's a pretty significant proportion of the time.

@grimalkina@mastodon.social Yeah, I can easily imagine that. I've certainly seen that dynamic elsewhere and have been on the receiving end of a few angry rants at work just because someone was frustrated by some unrelated situation.

@tobinbaker Generally, if someone says that something is "for X" I would take that to mean "intended for" or "designed for" which implies the intent of some person (or thinking entity). So I would say that it does imply design or foresight. This is, I believe, that generally if a biologist means that a trait confers an evolutionary advantage they say it is an "adaptive" trait (which doesn't imply it is the result of any intent or design).. Now sometimes people will say that a trait is "for X" as a more colloquial phrasing, of course, if they believe that the listener will understand that this is only a figure of speech. The irony is just that I would have thought Dawkins was precisely the sort of person who would assiduously avoid using any phrasing that appears to imply intent. @johncarlosbaez

@zwol Right. On a computer my feeling is that I should indicate the semantics (a period followed by whitespace to indicate the end of the sentence), and then the layout should be determined separately (possibly with user-defined preferences in a browser). I know LaTeX will ignore multiple spaces after a period, and I actually thought HTML rendering generally ignored multiple spaces as well (unless you use the non-breaking space HTML entity or are inside a pre tag). @luca @draNgNon @brad @jonny

@synlogic4242 @shac One example of a non-evil algorithm that I find very useful on Bluesky is the "Quiet Posters" feed, which just highlights recent posts from people you follow who don't post very often (so their content tends to get lost the shuffle of more prolific posters in the following feed).

@dougmerritt@mathstodon.xyz You said: "But as practitioners, we must assume at least for the moment a number of pragmatic concerns, such as that reality is real, in some sense." But don't interpretations like QBism sort of deny exactly that? https://en.wikipedia.org/wiki/QBism @davidsuculum@mathstodon.xyz

@vacapinta@mastodon.coffee A nice fedi thread on it here: https://spore.social/@minouette/114602418316375119

@grimalkina@mastodon.social I think you accidentally wrote "randos" when you should have written "idiots." 😉

@nix I also think that your point about human propensities is very well put. When we speak to other people we very seldom if ever explicitly state all relevant details (because to do so would be completely impractical). We assume that the listener will use shared context to infer all the additional details. So we are generally listening to what people say, assuming they have some idea behind it that makes sense, and then extrapolating their exact meaning. This is precisely why it's possible for people to fake their way through conversations in some cases: as long as they know the sorts of things they're expected to be saying and the listener expects to hear something sensible, they will extrapolate what is said to some coherent idea (within some bounds). Add to this the human tendencies toward pareidolia and anthropomorphism in general and it seems like we are very poorly adapted to detect whether an AI system "understands" what it's talking about or has human-like cognition in other respects by reading human language that it outputs. And on the other side, it seems LLMs are constructed in such a way that they should be specifically good at passing this sort of test even if they were not capable of understanding or other human-like cognition, because their base mechanism encodes and reproduces the structure of human language specifically. Having encoded information about virtually all text in existence, it would be very hard to tell if you're getting regurgitation of a compressed representation of the training data containing human thought or truly novel thought produced endogenously by the AI. @johncarlosbaez

@nix It would seem like if you want to test the putative cognition of an LLM you would specifically want to find a method where producing human language is not a large factor; for example, you could give it an interface that allows it to play a game or manipulate a robot to solve problems (ideally not ones specifically described in common texts). One example along that line was this attempt to evaluate the ability of coding assistant LLMs to code in programing languages with very idosyncratic textual representations: https://arxiv.org/abs/2603.09678 @johncarlosbaez

@idlestate I guess LaTeX just solves the problem by putting the onus on writers to indicate an abbreviation differently than a full stop at the end of a sentence. That's probably not a realistic approach for something like HMTL, where most of the writing is informal. @zwol @luca @draNgNon @brad @jonny

@davidsuculum@mathstodon.xyz That's a good point! You're correct that there are a number of such theories. The one I've encountered quite often is the Diósi-Penrose model. I do think of these as being somewhat in their own category, though, precisely because they actually propose to make an empirically-testable modification of the Schödinger equation in order to address things like the measurement problem. @androcat@toot.cat @dougmerritt@mathstodon.xyz

@dougmerritt@mathstodon.xyz I don't know all that much about it, but see, for example, the discussion of Lorentz invariance here: https://plato.stanford.edu/entries/qm-bohm/ "Bohmian mechanics is not compatible with special relativity, a central principle of physics: Bohmian mechanics is not Lorentz invariant. Nor can it easily be modified to accommodate Lorentz invariance. Configurations, defined by the simultaneous positions of all particles, play too crucial a role in its formulation, with the guiding equation defining an evolution on configuration space." But, as it says there, there are modifications that purport to be Lorentz invariant. @davidsuculum@mathstodon.xyz @androcat@toot.cat

@androcat@toot.cat But I honestly view a certain version of the "orthodox" IQM this way, as being demonstrably wrong.

By this I mean the view that macroscopic measuring apparatuses interacting with microscopic quantum systems induce a non-deterministic, irreversible state reduction, a change in the actual state of the system rather than just our knowledge of or belief about the system. There are, of course, many adherents of the "Copenhagen interpretation" who would say they don't take this view, but in my experience this is exactly how many working physicists view it.

This is definitively incorrect, because it's inconsistent (without additional ingredients): macroscopic systems comprise many microscopic quantum systems, and a collection of quantum systems still obeys QM, with deterministic, reversible dynamics.

This is precisely why when I want to think carefully about any subtle aspect of quantum measurement I turn to MWI, since it at least seems to present a self-consistent way to understand what's going on*. Granted, for most practical purposes thinking in the (ultimately incorrect) way outlined above works just fine.

• This is not to say MWI is the only self-consistent IQM, it's just the one I'm most familiar with, outside of the spectrum one might refer to as "orthodox" or Copenhagen interpretation.

@dougmerritt@mathstodon.xyz @davidsuculum@mathstodon.xyz

@androcat@toot.cat I guess I'd roughly think of interpretations of quantum mechanics (IQMs) as marrying a certain ontology to the math and observed behavior of quantum mechanics (QM). So it's definitely true that these can rule in or out certain statements about ontology, but they always have some story for how it still matches up with the math and observations. I'd also point out that while we can expect there to be many IQMs that are consistent with the math and observations of QM, I think it is possible to have an IQM that you discover is actually inconsistent with some of these fixed aspects of QM. The simplest example of these would seem to be pilot wave theory, which in its most rudimentary form is not, so far as I know, consistent with relativity. So while it does match non-relativistic QM (which makes it quite interesting), it does not extend to relativistic QM (though I gather that there are extensions or modifications of pilot wave theory that purport to). @dougmerritt@mathstodon.xyz @davidsuculum@mathstodon.xyz

@dougmerritt@mathstodon.xyz "Sorry if I'm being confusing about what position I'm taking, but it's because at the moment, I'm not taking a position." Or perhaps you're taking a ... superposition. 😜 More seriously, though, I generally agree that even if we can't definitively prove the objective existence of things, it's usually the most practical working assumption. So I personally consider myself a realist/materialist,

@dougmerritt@mathstodon.xyz But in a more skeptical view one might argue that we have direct, incontrovertible access to very little information about the world, only our perceptions. Even an object that you see in front of you can be subject to optical illusions, or you could be hallucinating, etc. In this framing, one can only mostly make statements of belief or credence about the world. My point being mostly that even trying to define the scope to which to apply the two different concepts of probability is pretty difficult. @davidsuculum@mathstodon.xyz

@davidsuculum@mathstodon.xyz Interesting. For the first point, it seems this could mean the quantitative problem from deriving the Born rule or the more philosophical problem of assigning probability in a world where every possible outcome of a measurement happens (in some branch of the wavefunction). Since I think Carroll proposes to solve the latter by invoking a Bayesian approach in terms of credence, I suspect you're right that he must mean the former. Personally, while I don't object to the Bayesian approach to statistics in general, I do find it vaguely dissatisfying to turn an empirical prediction (as in the usual frequentist formulation of the Copenhagen interpretation) into a statement about belief. For the second point, I agree with you; I don't understand why this should be a problem particular to MWI. Presumably in MWI, as in any other interpretation, classicality must arise from decoherence, and I would expect the problems there (proving that it's a generic phenomenon that will lead to the expected classical dynamics) to be the same. He has probably already expounded upon this at length, but I'm not sure where. It would be interesting to read.

@qgustavor@urusai.social The news remains relevant, and many people likely didn't catch it when it was new. That being said, I agree that we should make it the norm to link to a reputable source when passing on important information. Sharing ostensible screenshots from social media is how a lot of misinformation is spread. @maccruiskeen@social.linux.pizza @VeroniqueB99@mastodon.social

@spacebug@n4.mikronod.se IVPN is one of the few VPNs recommended @privacyguides@mastodon.neat.computerhttps://www.privacyguides.org/en/vpn/

@aumalatj@mementomori.social @mrundkvist@archaeo.social Being trained as a scientist but raised as an American generally leaves one with a strange relationship to units. I would always do calculations in SI, almost all the physical constants I know are in SI, and I wouldn't even know some of the unit conversions in US customary units or, in some cases, even how the units are defined for some quantities (e.g., what's the equivalent of a tesla in E&M?). But all my formative experiences were in US customary units; I have a sense of how a certain temperature in Fahrenheit feels. I have an intuitive sense of how long a mile is or how big an acre is. So I would always compute things in SI and preferentially use it for any precision task, but if I want an intuitive sense of a quantity*, it can be useful to convert it to US customary units.

• of course this only applies to quantities that fall within the regime of direct human experience. Since the two areas I spent the most time doing research in were quantum physics and astrophysics, this didn't come up all that much. 😁

@tuban_muzuru This criticism seems rather misplaced. @jamie made an observation based on the text of the CRS report, and he's clearly not afraid of LLM code generation, indeed he uses it. You can reasonably disagree with his interpretation, but you didn't make a very convincing argument here and instead started criticizing things you assumed about him and his attitudes rather than the substance of what he said. I also tire of the uninformed hyperbolic criticism of LLMs that one often sees on Fedi, but this really isn't that. It seems like maybe you've let your frustration with that bleed through to a discussion where it is not really relevant.

@baldur@toot.cafe This seems to implicitly assume that the demand for software will be constant, but conventionally we'd expect it to go up when the price goes down. Whether this works out well or poorly for software developers would then depend on how much demand goes up (i.e. the elasticity of demand). I think that applies to the cases in which skilled developers are still needed but the tools increase productivity. In the hypothetical case where AI coding tools are so effective that developer skill becomes completely irrelevant it seems like you couldn't have a functional market.

@setiinstitute@mastodon.social Damn, and here I thought @adamshostack@infosec.exchange had gotten a cool new nickname.

@zwol 😆

@spacebug@n4.mikronod.se Oh, and @ivpn@mastodon.social is also on Mastodon.

@aumalatj@mementomori.social @mrundkvist@archaeo.social This is a mistake that people often make in consuming science communication, assuming that the communicators speak in colloquial or simplified terms because they lack an understanding of the subject matter, when in actuality they are often simply phrasing it in a way they think is most likely to be understood by their target audience. And the target audience is almost never experts nor usually even enthusiastic amateurs. In this case it might be the American of median education level or perhaps the median middle school student (if trying to get kids into the science education pipeline).

@joshsutphin@wandering.shop @xgranade@wandering.shop Yeah, a while back I posted about how I see it as an inherent danger whenever some appliance has an associated app. https://mathstodon.xyz/@internic/114440391897810557

@sundogplanets@mastodon.social Oder, für die Deutschen, goaten Morgen. 😉