sparksbet's recent activity

  1. Comment on An AI social coach is teaching empathy to people with autism in ~health.mental

    sparksbet
    Link Parent
    This is very true -- writing is a technology external to the actual meat of human language -- but I don't think the difference is super relevant to many of the skills involved here. Learning the...

    Learning to write in a foreign language and being able to hold a conversation in that language are not entirely unrelated, but they involve different skills that both require practice.

    This is very true -- writing is a technology external to the actual meat of human language -- but I don't think the difference is super relevant to many of the skills involved here. Learning the right thing to say to seem empathetic to a neurotypical person is definitely something that can transfer between spoken and written contexts to at least some degree.

    Anecdotally, as someone with a recent autism diagnosis, I do find written communication a lot easier than spoken communication, in large part because there's a bit more time to respond even in highly synchronous textual communication, which lessens my anxiety a lot. But I both have social anxiety and have never really been considered unempathetic even in spoken contexts, so I've probably not got the same flavor of symptoms as the autistic people who benefit most from this type of coaching. So, y'know, grain of salt.

    6 votes
  2. Comment on Therapists are secretly using ChatGPT in ~health.mental

    sparksbet
    (edited )
    Link Parent
    We do not have remotely the evidence that it is something that can be measured in any objective way, including with future technologies, because there is nothing even close to sufficient evidence...

    There's nothing logically stopping a objective measurement of a person's brain state and formally determining whether someone is or is not depressed.

    We just don't currently understand the human brain well enough or have the technology to accurately measure its state. Its not technically impossible, it's just completely infeasible with where we currently are in neuroscience.

    We do not have remotely the evidence that it is something that can be measured in any objective way, including with future technologies, because there is nothing even close to sufficient evidence that depression would be directly observable from the state of one's brain. Currently the evidence I'm aware of points to the symptoms we classify as depression not necessarily all being caused by the same sources physiologically -- different people with depression could have completely different things causing it underlyingly. Heck, that's already obviously true to some extent -- I had depression that was caused at least in part by my thyroid not working properly.

    It's not impossible that eventually we'll develop the ability to directly test for and measure some mental health problems through neuroscientific methods. But it's also not remotely unlikely that for many mental health conditions, doing so is indeed not possible.

    3 votes
  3. Comment on Call of Duty: Ghosts – Power, paranoia, and orbital tungsten rods in ~games

    sparksbet
    Link Parent
    This is the other talk I watched and it was really funny lol! I doubt it'll be getting its own video like the Call of Duty: Ghosts one, but imo it was also much more fun to watch the recording.

    This is the other talk I watched and it was really funny lol! I doubt it'll be getting its own video like the Call of Duty: Ghosts one, but imo it was also much more fun to watch the recording.

    1 vote
  4. Comment on Call of Duty: Ghosts – Power, paranoia, and orbital tungsten rods in ~games

    sparksbet
    Link Parent
    idk, Why It's Rude to Suck At Warcraft may have technically come out a couple years ago, but it's one of his more recent videos given the pace at which he uploads (it came out after Line Goes Up)....

    he moved on from gaming content years ago

    idk, Why It's Rude to Suck At Warcraft may have technically come out a couple years ago, but it's one of his more recent videos given the pace at which he uploads (it came out after Line Goes Up). And that's if you don't count the Decentraland content as being about gaming. I wouldn't characterize him as having moved on from gaming content at all; it just hasn't been the focus of his last four or five videos.

    He gave two recent PAX talks that I've seen recordings of, and iirc one of them was about Call of Duty: Ghosts, so I'm interested in seeing how he's refined it for a video!

    6 votes
  5. Comment on I had an idea for a Crusader Kings, but about rich families in Victoria-Modern Era. What could go wrong? in ~games

    sparksbet
    Link Parent
    I think in terms of how the systems you build tell stories through the incentives they create, there's a lot of interesting stuff you can do there. I think having incentives for the...

    I think in terms of how the systems you build tell stories through the incentives they create, there's a lot of interesting stuff you can do there. I think having incentives for the player/characters to do things that were bad in real life isn't necessarily a bad thing -- you just need ro be very conscious of the story those incentives are telling about real-world history. This can be more complex in a simulation game like this than it is in a lot of other types of game.

    You mention slavery, so I must link to this video that goes over how various Paradox games handle slavery. It's relatively short and iirc very good in how it analyzes the way different Paradox games treat the issue, and there are great comments analyzing the issue further too. It comes from the perspective of a Paradox fan as well, fwiw. He also goes through a lot of interesting history as context as well! The section on Victoria 3 is particularly apt imo.

    Definitely mostly a design question for once you've got the basics up and running, as opposed to this stage in development. But I encourage you to explore what your mechanics say about the history when making a game like this -- there's a lot to sink your teeth into there!

    1 vote
  6. Comment on McDonald’s is cutting prices of its combo meals to convince customers it’s affordable again in ~food

    sparksbet
    Link Parent
    The words "saturated" and "trans" do not appear in my comment or in the comment I replied to. The idea that someone "shouldn't want" them is absurd. Your body's hunger signals and cravings are not...

    The words "saturated" and "trans" do not appear in my comment or in the comment I replied to. The idea that someone "shouldn't want" them is absurd. Your body's hunger signals and cravings are not granularly distinguishing types of fats like that, and your body does actually need fats in general. Much like it also needs salt.

    Also, don't tell me what I should and shouldn't want. It's not your place.

    6 votes
  7. Comment on McDonald’s is cutting prices of its combo meals to convince customers it’s affordable again in ~food

    sparksbet
    Link Parent
    Sometimes your body does need fat and salt, though. Not always, of course. But insisting that certain things -- especially basic nutrients like these -- are universally bad and evil to eat doesn't...

    Sometimes your body does need fat and salt, though. Not always, of course. But insisting that certain things -- especially basic nutrients like these -- are universally bad and evil to eat doesn't actually help improve people's health. In fact, fucking up someone's relationship with food by moralizing too excessively is potentially far more harmful to their health, as it can absolutely itself lead to unhealthy or even disordered eating habits.

    6 votes
  8. Comment on Famous cognitive psychology experiments that failed to replicate in ~science

    sparksbet
    Link Parent
    I've been hornier when ovulating at times for sure (though it's far from as consistent as a lot of people make it out to be), but I haven't noticed it changing who I am attracted to on any axis.

    I thought that women just find every guy attractive when they're ovulating.

    I've been hornier when ovulating at times for sure (though it's far from as consistent as a lot of people make it out to be), but I haven't noticed it changing who I am attracted to on any axis.

    4 votes
  9. Comment on Thoughts on wallpaper? in ~life.home_improvement

    sparksbet
    Link
    We removed wallpaper from one of the bedrooms in my childhood home -- wallpaper that was definitely not remotely as old as the 60s to 80s... it was from the 90s at the oldest, and it was an...

    We removed wallpaper from one of the bedrooms in my childhood home -- wallpaper that was definitely not remotely as old as the 60s to 80s... it was from the 90s at the oldest, and it was an absolute ordeal that took at least a week of full-time work from multiple people (though admittedly all family members rather than professionals). I think you are underselling how bad removal can be.

    Aesthetically, wallpapering all four walls often doesn't fit my sensibilities, but there's so much variance there based on personal taste that it's harder to weigh in there.

    5 votes
  10. Comment on What heritability actually means in ~science

    sparksbet
    Link Parent
    Ah, yeah, that's a fair enough observation. It's a common thing to hedge a little when making claims, just in case you happen to be mistaken or have gotten something wrong (I know I do this a lot...

    Ah, yeah, that's a fair enough observation. It's a common thing to hedge a little when making claims, just in case you happen to be mistaken or have gotten something wrong (I know I do this a lot myself), so it's refreshing, straightforward writing to not hedge at all and just say "I'm right." And in this case, I think the author backs it up enough that it pays off!

    5 votes
  11. Comment on No comply: private equity and skateboarding, or how private equity is gutting skateboard culture in ~finance

    sparksbet
    Link Parent
    The vast majority of people who argue for a return to the gold standard are weird crackpots (or fans thereof parroting them), so I'd take it with a pretty big grain of salt.

    The vast majority of people who argue for a return to the gold standard are weird crackpots (or fans thereof parroting them), so I'd take it with a pretty big grain of salt.

    4 votes
  12. Comment on What heritability actually means in ~science

    sparksbet
    Link Parent
    Not really sure why, tbh. The LLM is responding in accordance with common human assumptions/misconceptions in the same scenario -- I too immediately assumed English's heritability would likewise...

    Not really sure why, tbh. The LLM is responding in accordance with common human assumptions/misconceptions in the same scenario -- I too immediately assumed English's heritability would likewise be zero at first because I didn't take second language learning into account.

    6 votes
  13. Comment on ‘Being short is a curse’: the men paying thousands to get their legs broken – and lengthened in ~life.men

    sparksbet
    Link Parent
    I mean that's just something people actually say all the time lol

    I mean that's just something people actually say all the time lol

    3 votes
  14. Comment on What's a question you could ask to determine if someone is an expert in your line of work? in ~talk

    sparksbet
    Link Parent
    For undergrad, I attended one of the few US universities (perhaps the only one?) with a non-Minimalist focus when it came to syntax. I still remember how much fun it was when they invited a...

    For undergrad, I attended one of the few US universities (perhaps the only one?) with a non-Minimalist focus when it came to syntax. I still remember how much fun it was when they invited a Minimalist from another university to give a talk -- I did formal semantics/pragmatics, so I didn't really care much about the finer points, but no one bickers like the syntax guys.

    3 votes
  15. Comment on What's a question you could ask to determine if someone is an expert in your line of work? in ~talk

    sparksbet
    Link Parent
    God asking someone if they're a generativist is such a boringly broad question to ask at the theoretical linguistics parties. At least ask about UG to get a real fight going.

    God asking someone if they're a generativist is such a boringly broad question to ask at the theoretical linguistics parties. At least ask about UG to get a real fight going.

    8 votes
  16. Comment on Germany legal case alleging adblockers violate copyright in ~tech

    sparksbet
    Link Parent
    The German legal system is notably not a common law system, but a civil law system. One of the principle differences is that German courts are not bound by the precedent of previous court decisions.

    I don't know how the German legal system works, but the US legal system is common law, meaning the law is heavily defined by precedent.

    The German legal system is notably not a common law system, but a civil law system. One of the principle differences is that German courts are not bound by the precedent of previous court decisions.

    25 votes
  17. Comment on ‘Sex reversal’ is surprisingly common in birds, new study suggests in ~science

    sparksbet
    Link Parent
    It took me an embarrassingly long time to realize this was a Jurassic Park reference lol

    It took me an embarrassingly long time to realize this was a Jurassic Park reference lol

    6 votes
  18. Comment on Claude Opus 4 and 4.1 can now end a rare subset of conversations in ~tech

    sparksbet
    Link Parent
    Most of your responses don't require any response from me, but a couple quick things: The short answer is no. Even if we had perfect access to tons of actual, non-reconstructed data from something...

    Most of your responses don't require any response from me, but a couple quick things:

    Would you consider an analysis of how latent space manifolds change over a variety of LLMs restricted to historical language to be relevant data (admittedly hard to find)?

    The short answer is no. Even if we had perfect access to tons of actual, non-reconstructed data from something like PIE (which isn't just hard to find, but straight-up does not exist, because PIE is a reconstruction of a prehistoric language), that language would not prove anything about the origins of human language unless you could prove that it itself constituted the first human language. Even our oldest reconstructions of prehistoric languages are indubitably not even close to the first human languages -- we stop trying to reconstruct further back not because these languages don't themselves have ancestors, but because we lack the data to do so (and aren't crackpots). The remoteness of any data about the origins of language really cannot be sufficiently exaggerated -- anyone who isn't a complete quack would acknowledge that it's difficult to even extrapolate about it.

    my point is that even without direct access to the outside world an LLM has some cognitive parallels to it thru language as a lossy fossilized embodiment of the external world, thus indicating the comparison to an animal has some validity

    I'm sorry, but this is simply nonsense. An LLM having representations of language that correspond in some systematic way to the human brain's responses to visual stimuli does not remotely indicate that comparison to an animal has any validity. Results like those in this paper are fascinating from a neurological and psycholinguistic standpoint, and the potential implications about how language itself reflects human cognition about the world is interesting. But this is so wholly removed from anything that even approaches what we would call sentience or that would merit comparison to an animal or human that doing so on this basis indicates a fundamental failure to understand what language is and what these results actually show.

  19. Comment on Claude Opus 4 and 4.1 can now end a rare subset of conversations in ~tech

    sparksbet
    (edited )
    Link Parent
    So, for context, I have an academic background in theoretical linguistics and professional experience with NLP and LLMs. Just putting that up-front to establish my perspective there. I'm not...

    So, for context, I have an academic background in theoretical linguistics and professional experience with NLP and LLMs. Just putting that up-front to establish my perspective there.

    I'm not really sure what distinction you're drawing between "the meat of language" and "the formally consistent and thus empty structure" is -- I don't think that a formally consistent model of language would be "empty structure" or would be distinguishable from "the meat of language" in the way you describe even under even the most formal approaches to theoretical linguistics. If it's possible to build a formally consistent model of language -- and many theoretical linguists do believe it is -- such a model would not be "empty structure" and would include the "meat of language," whatever that is in your definition.

    I think any claims that something definitely constitutes the reason for the evolution of language in humans should be taken with a huge grain of salt, since there's simply very little evidence to rely on here and thus the claims are very vibes-based and tenuous in the same way as a lot of evopsych -- a huge portion of the claims there are just unsubstantiated quackery (I would include the concept of the bicameral mind in that category, so far as I understand the current scientific consensus and the state of the evidence for or against it).

    Due to the lack of actual data to study when it comes to the origins of language, it's not something I have much direct experience with from an academic/theoretical perspective. Obviously there are connections that can be made between various types of animal communication and human language, and there's evidence that the advent of language corresponds with large increases in human cognitive abilities. But a lot of the more specific claims about why the sheer complexity of human language developed have no more supporting evidence than the Tower of Babbel does.

    Unless you take a very broad view of what constitutes "describing a shared environment", I think that describes only a fraction of the utility of language. Language accomplishes a huge number of functions beyond that, and if I were to make my own vibes-based guess, I suspect that other speech acts are more "fundamental" than the description of one's environment. I'm not convinced that describing a shared environment is more fundamental to the evolution of language than making requests or any number of other speech acts. But I'll confess that I simply don't find the ultimate origins of human language all that interesting, and like the majority of linguists I'm far more interested in describing language as it can be observed and studied today than speculating on such matters without evidence.

    Moreover, even accepting "describing a shared environment to other beings" as a motivating factor for the evolution of human language, that is not comparable to how LLMs work. Reading the paper you linked (which is super interesting fwiw), it itself doesn't draw that comparison or make any claims about the evolution of human language. Their results seem to show that there are systematic correspondences between the LLMs' underlying representations of the text of image captions and the neural responses humans have to the images as visual stimuli, to the extent that models can be trained to predict the latter from the former. This is fascinating, but it has absolutely nothing to do with the "sentience" of LLMs or the origins of language. The LLM representations of image captions were certainly not developed under anything remotely similar to the conditions under which human language developed or by any method that even approaches "trying to describe their environment to other beings". On the contrary, the LLMs used text-only embeddings and had no access to the visual context of the images whatsoever. This is obviously wildly different from how humans, modern or otherwise, learn the correspondences between language and their environment.

    Which is part of why the results of this paper are interesting in the first place! It would actually be a much less interesting result if the LLM were a multi-modal model that had access to the visual representations of the images and was modeling them directly. But the idea that this paper serves as evidence for even similarities between how LLMs learn and how human language developed, much less as evidence for anything approaching sentience or consciousness in LLMs, is not supported by this article and is a pretty absurd leap to take. Even the article's speculation on how these techniques could potentially enable communication with other organisms (a fascinating but remote prospect imo) doesn't go even close to that far, and even if their wildest suggestions there materialize, it still does not prove anything about the actual evolution of human language.

    Even if they'd used a multi-modal model that did have access to visual information along with the text (and thus corresponded better to the concept of "describing a shared environment to other beings") and the result was some sort of systematic correspondence between human brain activity and the underlying features in the LLM when it came to the linguistic representations themselves, that still would not have been evidence that the way both these systems developed is the same, because it is absolutely possible (and, in fact, quite likely) that extremely different processes could result in very similar systems. Convergent evolution is a thing. And that's even if we set aside the fact that no LLM is ever making a language "from scratch" and is learning from existing linguistic data (and huge quantities thereof) that definitionally would not exist prior to the evolution of human language. But this is all so disconnected from anything actually in the article you linked that it's more or less moot anyway.

    1 vote
  20. Comment on Claude Opus 4 and 4.1 can now end a rare subset of conversations in ~tech

    sparksbet
    Link Parent
    This is overstating our knowledge of how human language developed to the extent that it's misinformation. A single paper claiming that language may have evolved this way does not establish this as...

    consider the fact that language itself evolved out of humans trying to represent embodied cognition

    This is overstating our knowledge of how human language developed to the extent that it's misinformation. A single paper claiming that language may have evolved this way does not establish this as fact even if you take all its claims at face value, and the idea that this is how language developed in humans is not remotely within the ballpark of being well-established, much less being scientific consensus.

    4 votes