kunai

Definitely part of the issue (and the vast majority of people who ignore it because they're too PC to call it out are as well) however it's a small piece of the total pie. There are other bigger economic choices that are worse and causing more immediate harm

I want to be clear: I hate this. I hate everything about it. HOWEVER...

If you want to argue against wetware then we need a better argument than this. The same ideas were used to argue against stem cell research, IVF, and many other treatments and ideas that have saved and helped lives. Unit 731 isn't an apt comparison because cells are not inherently sentient to our knowledge. If they were it would absolutely shatter the profession of medical ethics as we know it.

Oh don't get me wrong, I concede I could be totally wrong. But there are other scenarios that are apocalyptic that aren't model collapse and they are equally as terrible. Civilization lives and dies on our edge cases.

> Why does everyone value themselves based on their economic output?

I mean this is really the issue with the idea of AI taking jobs, isn't it? If our entire economic system wasn't predicated on this, nobody would be really that miffed about AI use in art because there wouldn't be much of a market for it, because the concept of art as being a marketable widget wouldn't exist in the first place, and the notion of human expression being replaced with a machine would be absurd. But we have tied up art with economic exchange for centuries. It has been a market commodity for almost as long as recorded human history. So the graphic designer who pours their creative spirit into what is, essentially, an economic exchange (rather than going and painting landscapes or something) feels an acute threat because it threatens their existence.

To overcome this and to live a future where it feels more like ML/LLMs live in concert with humanity then you'd have to picture a future where human value is divorced from one's capacity to provide economic value. When exactly does anyone foresee that happening? The entire system would crumble. UBI is clearly not the answer because all it does is entrench a subtle cast system and limit upward mobility and does nothing to address this root cause.

I don't know what the future is, but I know that as long as the current economic system persists (in all its permutations, whether that's socialism, capitalism, communism, whatever outmoded 19th and 20th century ideas we have to structure society -- all alienate the individual and relationships to some degree or another) then AI will always be met with disdain even in areas where it can improve people's well-being (and FWIW I think the arts and even mathematics are NOT part of this subset of use cases).

Maybe the solution is walled cities and shutting off the internet entirely; returning to our neighborhood pubs, sports, expressions of the human spirit placed up at farmers' markets. Seems like we already seem to be trending in that direction; and if last night's apocalypse was any indication maybe the internet's days are numbered anyways.

> a trend of ai winning awards is either indicative of the quality of the writing or of a systemic flaw in how these awards are judged (which seems plausible too).

I definitely think this is plausible, and I think we need to start extending the idea of code "smells" to regular languages as well. I have TAed for literature and have reviewed many a manuscript and except for really really finely honed projects with lots of editing in post, written text almost always has many of these obvious smells (cliches, fallacies, illogical juxtapositions, poor cause and effect). AI-generated prose is no exception. Character names, phrases, descriptions of sensory experiences that are off or not quite right, improper negation (LLMs still struggles with the idea of 'scope'), sentence structure being very predictable and one-dimensional (even when using the latest models and mimicking famous authors and poets and even with very careful prompting), so on and so forth, you get the idea.

> A perfect replica of a van gogh painting conveys the same artistic expression as the original.

I think OP's example was a poor one. A perfect replica of a van gogh painting indeed does convey the same artistic expression as the original. However, I think the better comparison is an understudy's mimickry of A Starry Night which seems inspired by it and yet adds nothing to the original and is somewhat deficient in some ways, and misunderstands the artistic intent of the original. That would certainly not convey the same expression as the original, and would be met with widespread criticism.

They are delusional in ways that don't overlap, though. An LLM will not be delusional in the way Van Gogh or Howard Hughes was, but humans (sane ones anyways) will not hallucinate the exact kind of odd, unpredictable LLM output that makes you go "huh? why the heck did it do this in this way or write this." Things which are extremely obvious well within the limits of even elementary human reasoning are completely outside the purview of model capability; it's much easier for a model to "reason" when it comes to systems and ideas that are very discrete (which is why it's so useful when writing code, esp. in higher-level languages -- little uncertainty there and a round-robin verification process for workflows like mathematics. However once you ask it to intuit anything that falls apart quickly; its main utility lies in being able to compensate for these deficiencies with sheer volume and burning through tokens)

Even on the frontier models these blips still happen fairly often and it's intrusive to a lot of workflows.

Looks somewhat vibe-coded. How much of this is independent work? If it's actually faithful graphic design and integrated, well-thought out code with a bit of Claude glue on top then I'm fairly intrigued, but if the entire thing was built with llm tooling then that does unfortunately sap my interest in this a lot.

I do love the idea though. I do also hate the fact that if I saw a project like this six or seven years ago I'd be in awe at it, but now that there's far less effort required, I just go, "oh neat, eh, but how functional is it really" and move on.

Learning what infinitesimals were and how useful they were for understanding the rates of change was an actual eureka moment for me, and many people in my classes felt similarly. It fundamentally changes the way you view things as simple as a car accelerating or the way heat spreads through a system, the way fruits spoil faster in the winter if you have heat, why metals creak when they're cold and why they groan when they're heated.

The thing that really begets me in all of this is that what you describe, working hard pursuing a skilled passion, requires a foundation of knowledge that up until a year ago was considered a foregone conclusion for becoming someone of note. It doesn't matter whether someone learned it autodidactically or through structured schooling, the point is that the knowledge itself is fundamentally important. We never stopped teaching children basic arithmetic or mental math when slide rules, adding machines, and calculators became widespread. The reason why is pretty obvious.

And yet, the technofeudalists want us to stop teaching our children critical thinking, science, arts, technology, and everything else that encompasses the indomitable human spirit, and why? Because a machine can do those things too? Machines have been capable of or better at a great many things that humans are also capable of for the past hundred years or so and yet we have never stopped teaching humans those very things exactly because we used to recognize that those things had enormous utility in general in a sphere far beyond whatever is necessary for your calling or profession. It is absolute lunacy to me that someone could see LLMs and agentic flows do brute-force proofs and then somehow suggest that the entire field of mathematics is obsolete.

I urge you to read the piece "Complex Systems Won't Survive the Competency Crisis." If we lose all of our institutional knowledge of how the puzzle pieces all fit together there will be no one to save us when (when, not if) the systems we're creating now suffer a catastrophic failure. And they will. Model collapse has not happened yet because of luck. Synthetic data can only get you so far. And whenever we enter our Idiocracy/WALL-E stage of total compliance and deference to machines themselves, one day there will come a time when something breaks and no one knows how to fix it because they no longer understand the underlying principles.

You might say in an optimistic scenario you can just ask the LLM for the tools and resources and prior knowledge needed to fix it. Alas in many cases, this is barely enough, and most on this forum are technically knowledgeable enough to actually make use of that advice. It's hard to see a compliant, totally intellectually decrepit population doing the same. I'd also advise you that although unschooling has its benefits, it's a movement that is filled by those who are often cultists and other fringe hacks who have grievances with the school system for reasons unbeknownst to me and many others. I would not use them as any sort of advice forum or channel for meaning, and the fact that you seem to be taking them as gospel is extremely questionable and makes me uneasy about the future of this forum in general.

If you've been in the military (or close to it, as I have) you would not peg the top brass as actual intellectuals or geniuses. They are very easily convinced with buzzwords and fluff that is reflective of their inner emptiness. There are exceptions, of course (like Scott/Mark Kelly) though they are few and far between.

This was Air Force. YMMV for other branches.

I genuinely think the frontier labs are just throwing shit at the wall to see what sticks. All of the improvements since Fable 5 seem marginal. Most of the quoted AGI "risk" is coming from super large agentic models that are juggernauts that can brute force their way through any problem faster than a human can. Despite that they're no closer to AGI than they were a year ago; the technology's limitations are apparent the more you use it. Doesn't mean it's still not extremely dangerous and/or capable, it's just not the "replace a human in 99% of cases" thing. There are obvious tradeoffs besides just the models being faster and capable of information retrieval and generation at a rate exponentially faster than a human's.

My long-term hope (and call me another Zitron if you wish) is that the labs' hype dissipates if there's no serious improvements beyond stringing together agents to ram through brick wall Millennium Prize style problems and we all just use on-device AI for 99% of cases where it's useful, researchers, governments, militaries, and universities can pay for the more complex models, and image/video gen dies a slow death (if Congress will actually legislate and/or SCOTUS decides that training those specific models does not qualify as fair use unlike training LLMs) and cost for compute rises over time as public interest in the tech sours.

LLMs are such marvelous and incredible technology squandered by genuinely deranged Silicon Valley cultists who are trying to use them as a Trojan horse to force their antisocial visions of the future upon an unsuspecting populace. We could all have just invested in on-device compute and nobody would have lost their jobs in pointless layoffs, productivity would have increased, and maybe we would have forced a conversation about when and when not to use AI and it wouldn't be so ever-present in use cases where it actually does harm to the consumer and society. But sama, PT, and dario just wouldn't have it that way, would they? Because if that were the case, there's no prospective hope where they can assuage their deep-seated insecurities over being antisocial and off-putting by reassuring themselves that there's an imminent realignment of society where they will hold all the cards when the dust settles.

Very intriguing, brings to mind Sapir-Whorf a little bit. You wouldn't happen to remember the name of the speaker or the conference, would you?

> Some of it is information you never had before, but are able to easily validate.

This is exactly the use case an LLM might (huge emphasis on might, depends on workflow, agentic vs. relying on contextual which can hallucinate) be good at and yet humans are notoriously bad at, because we are swayed by emotional responses and it is easy to have an emotional response to text that is programmed to look good for you and you alone.

> you communicate 1000 bits of information to me, instead of planned 500, and you yourself learn extra 500 bits.

Extremely optimistic. If this were the ideal scenario, you would USE the LLM to garner information ABOUT those 500 bits and then reframe them in a way that you yourself would put it. If there is insight, your "word" in your mental register now expands from the original 1000 bits to 1500 or 2000, and then are "processed" by your human brain that includes subconscious choices that are meaningful to the end result. There are tons of hidden semiotic data in your diction and wording (think resource forks in classic MacOS/HFS, only visible to the filesys) that is lost when you rely on another source to put together words for you; it's as if it is a game of Telephone. These are subtleties which you may intend for your recipient to receive and which are crucially important to your recipient and are irretrievable, it is intrinsically lossy. You have an alphabet soup of words, they cannot be put together by an LLM in exactly the way your brain did. We must rely on the fact that we ourselves put this together, the "aha" moment when an LLM does it for you is illusory and does not itself provide meaningfully important confirmation that you indeed say what you mean to say. Of course, humans say things and put things in way we do not intend to all the time. I still fundamentally believe this is more honest than relying on a third party that is not capable of understanding human emotional nuance to put together language for you, when language is and always has been a manner in which to dictate human emotional nuance.

> (Excuse the strong language, but I spent a lot of effort every day on both dealing with inconsiderate people lobbing LLM output at me, and making sure never to act like one myself, so it's a topic close to my heart.)

Does not negate the fact that LLM output itself, at least when used to convey human emotions or thoughts, is lossy. A very highly compressed JPEG with added interpolation from an upscale algorithm might come up with cool details that were never present in the original, and may look cool to both you and the recipient but are not honest to the source material. You receiving 240p JPEGs on a day-to-day basis is irrelevant to this. For the purposes of communication, it is a massive error which has the potential to compound, regardless of whether or not you or the recipient believe this to be the case.

I really would like to know what kind of drugs they place in the Bay Area water supply that produce comments like this.

Coward.