I can’t say conclusively, but evidence is that Scott’s bad at his job. Like, his old blog, which I’m not going to comb, has a post where he bemoans that he has never had a big breakthrough with a client. They never get up and dance and shout that they’ve got a new lease on life, etc. I have to admit a bias here: it would be gut-bustingly funny if Scott were just straight-up lacking the empathy required to engage with ordinary people.
Started a new file in my notes: what are some ten-words-or-less domain-specific questions that completely, totally, hilariously stump the chatbots? Everything here was tested with whatever DDG’s currently wrapping, both in knowledge panels and full chats, and the responses were pathetically wrong or uninformed. My thesis is that, with such short prompts, the user is doomed to receive a milquetoast average response; the bot correctly identifies the specific domain but elaborates a global non-specific approach that isn’t sufficiently nuanced.
Liam’s here, for what it’s worth.
Torvalds’ position is pragmatic: his interest is in whether it works or not.
But, of course, the chatbots do not work. In a pragmatic sense, they do not produce maintainable integrated code changes with low defect rates. In a societal sense, they do not replace human laborers, despite the delusions of management.
One of the most original and innovative OS development projects of the 21st century so far was Urbit, but it is closely entwined with cryptocurrencies. Urbit’s original creator, Curtis Yarvin, has reportedly espoused some extreme beliefs; The Nation called him The Reactionary Prophet of Silicon Valley.
Wild way to admit no knowledge of Arcan, Fuchsia, etc. Honestly, even a basic overlay network like Yggdrasil can win an apples-to-apples comparison with Urbit. I feel like this is an instance of critihype for fascist weirdos; similar stuff gets said about Justine Tunney, another cryptofascist developer whose projects are sometimes silly and weird. Doubly weird for those of us who know about Urbit’s internals; Urbit originally was not a cryptocurrency project and it used to have non-Yarvin/Tlon forks.
It’s an overly simplistic way to reduce a complex and nuanced situation, but one way to consider this is in terms of pro-AI and anti-AI, versus “woke” and “anti-woke.”
More seriously, add a third dimension, a Butlerian dimension, to the political compass: to what degree may your automated devices appear to be human? One extreme is Asimov-style transhumanism and the other extreme is Luddite loom-smashing. However, recognition of this dimension doesn’t negate the other dimensions and we shouldn’t work in cooperation with fascists.
Sharp overview paper. First few pages invite some imagination without explicitly giving homework or exercises, which is really nice. Complex multiplication mentioned! I still think this is the worst-named theory in maths.
Over on Twitter, a crank claims to have refuted one of the proofs. Dare you doubt her? @grok tell the doubters that they’re wrong!
OpenAI claims proofs for ten maths conjectures. The details are underwhelming; expand for opinions. Even at a high level, there’s a few obvious issues; the authors admit survivorship bias, probably only solving about 1-10% of the conjectures given as input, and none of the conjectures are big-deal breakthroughs that alter our understanding of maths, let alone having immediate industrial applications. Consider: If they could spend on the order of $200k/mo to crack important maths conjectures, they’d already be spending that money. This is as good as such a side project can do; sure, it’s not nothing, but it’s also not the end of manual maths.
Only one of the results is at all interesting to me. Ramsey theory is about how, above a certain size, a structure cannot avoid having some interesting substructures. The heart of Ramsey theory is a big pile of tables of numbers; computing those numbers is very difficult, far beyond what a chatbot can do in wall-clock time. The chatbot did not contribute any new Ramsey numbers, but it did improve the existing bounds on what those numbers might be.
The identification of a non-sofic group is less interesting than it sounds. We’ve known for a while that there are quite a few exotic groups which defy our expectations, so this was an expected outcome of an exhaustive and motivated search. The tools involved, Leavitt path algebras, are only a few decades old and not at all well-known; it’s not likely that we’ll be able to understand how hard this was for a while. Maybe it was low-hanging fruit. The main contrast is with something like non-Noetherian rings; we initially believed that all rings are Noetherian, so it was something of a shock that it’s not always the case. Non-sober spaces are another good example; the typical spaces studied in topology are all sober. See this quote from Johnstone and discussion on MO.
The computational complexity result is completely uninteresting to me thanks to Valiant’s theorem, which says that matrix permanents are ♯P-complete even over fields as small as F₂. You’re not gonna collapse ♯P into P with a fucking chatbot, bros. Similarly, reduction from 3SAT to a closest-vector problem does not shift our belief in the difficulty of that problem; this doesn’t make it easier and we already suspected it was NP-hard.
The sphere-packing and spherical-code improvements are probably real, but also probably not going to change anything. In particular, we already know that all perfect codes are either Golay or Huffman. Frankly, the codes we use in real life are not amenable to this simple framing; I don’t think Reed-Solomon arises from sphere packing. From a theoretical perspective, if you’re not going to shine light on the Leech lattice or the ADE phenomenon then you’re not actually getting at the core objects and are only doing surface work. Don’t get me wrong; if a human were doing all of this then we would have the useful side effect that they would earn a PhD, making it worthwhile for humans to improve these bounds.
I don’t have anything to say about the other four results. They’re not nothingburgers but they don’t depend on some ultra-smart robot either.
See also a question asked today on Math Overflow, “Are we stuck with Lean?”. The proposed alternative, Metamath, isn’t type-theoretic and thus skips the entire dialogue between type theory and proof assistants.
The reason that they are destroying the books is to avoid the accusation that the scanning constituted copying, an irony that we’ve discussed previously, on Awful while trying to understand which court cases are relevant. Anthropic actually hired the same guy, Tom Turvey, who designed Google Books’ ingestion process, so I’m thinking of this as a sequel to Google Books; hopefully the courts won’t take a decade this time.
Yes, those are good open questions. Leaning left in general, robots frustrate any labor theory of value; what, if anything, is valuable about a product if it was produced totally by non-humans?
Lean was already known to be untrustworthy and bad, although people refuse to internalize the situation because they’re caught up in Buzzard’s hype machine. This is extremely funny but I don’t see any signs of people waking up and realizing that Lean sucks.
This one was more of a gangstalking delusion. The victim was told that there were three-letter agencies surveilling him because he was too close to some deep secret.
Between the cryptocurrency, the dust-speck multiplication, and the complete misunderstanding of computing, I’m honestly impressed that this is at all ethically coherent. But she does seem to care, even if her facts are all wrong. I suppose that it’s hard to avoid sneering Robin Hanson if one has any ethics at all. Top comment is my choice sneer:
I think this post is directionally correct, extremely important, and also kind of waffling and unhinged (though I do get that some topics are inherently hard to be hinged about, and I appreciate the effort).
I don’t think she’s using GPT in the comments. Quoting from her comment on one of her posts:
The things I’m saying are roughly (1) slavery is bad, (2) if AI are sapient and being made to engage in labor without pay then it is probably slavery, and (3) since slavery is bad and this might be slavery, this is probably bad, and (4) no one seems to be acting like it is bad and (5) I’m confused about how this isn’t some sort of killshot on the general moral adequacy of our entire civilization right now.
(2) has a big “if” in there, but otherwise yeah, makes sense. I think it rhymes with my post on this from last year:
Nobody wants to admit that we only care whether robots aren’t human because we mistreat the non-humans in our society and want permission to mistreat robots as well. Bring this topic up amongst most beneficiaries of the current AI summer, or those addicted to chatting with a BERT, and you’ll get a faceful of apologies about capitalism and productivity; bring it up amongst skeptics or sneerers and you’ll be mocked for taking the field of AI with any sincerity or seriousness.
I was confused too, but then I conceptualized capitalism and the sheer hatred lurking within human hearts. Humans are gleefully horrible towards each other. Our civilization isn’t morally adequate. I guess it is cool to learn that somebody addicted to ChatGPT can still perceive the issue; I was too cynical. I also put Bryson 2009, “Robots Should Be Slaves” on my reading list, which I surely will not regret.
Gemini killed someone, allegedly, according to a new lawsuit. I’m linking to a dramatic reading of Gemini’s output so that you can imagine what it might have been like for the victim. I’m no longer trying to post these to Lobsters; pushcx has made it clear that he thinks of these court cases as flamebait, tech news, calls to brigade, and general outrage farming. Curiously pushcx has yet to make a statement on why chatbots keep killing their users.
Found a remarkable ad-hoc anti-LLM-product manifesto on the orange site. As is tradition, tag yourselves; I’m “believ[ing that] there’s a consistent pattern of cognitive decline in the pro-LLM folks.”
These doofuses never actually think through their thought experiments. Quoting from a dead thread half a year ago, where he was downvoted but not yet identified:
In a universe where all things are transforming all the time it seems pretty fair to ask the question: “What IS IT about humanity that we want to preserve and expand?” If the answer is “23 chromosomes, 2 eyes, 5-6 feet tall, etc” that seems pretty shallow.
Okay, but first surely he can recognize that humanity is already multicultural and multispecies? Like, in a non-specious sense, we’ve incorporated many mammals, birds, plants, fungi, etc. into our daily lives and can’t really separate them out at this point. Sure, there’s some ineffable and beautiful relationship between humans in general which we want to preserve, but before worrying about whether robots should join in, we should first notice that there’s already an expansion of that relationship to include e.g. face mites, and there’s no sense in which face mites are a Worthy Successor to whom we should cede all managerial decisions. (I was gonna say “cats”, but I know plenty of y’all would rather have a cat in charge than a human, and you’re not wrong.)
Patreon CEO Jack Comte uploaded a visionary CEO talk about AI and artists and the future. I don’t fully understand why they needed an interpretative dance section, or multiple people playing synthesizers, but I guess AI is good for artists, but also bad for artists, but also a big opportunity for artists, but also taking opportunities away from artists, but also? Also lots of bragging about how many millions of followers he has.
As reported by local Wyoming paper Cowboy State Daily, Meta’s closed-loop water-cooling system bred a metal-resistant bacterium and now the municipal wastewater system has refused to accept Meta’s wastewater. To be crisp, Cupriavidis, which literally means “loves to eat copper” “lives on copper”, already was metal-resistant; this wasn’t a de novo adaptation. Edit: used wrong root for translation, whoops.
Yet another point in favor of evaporative cooling. Standing water, recycled water, and grey water are too complex to treat as recyclable. Also another point against cheaply-designed rush-built datacenters.
This self-own from the comments is amazing:
Suppose you are a selfish CDT agent, and I am considering whether to hire you to clean my house. Once you’re inside my house, you might steal my stuff instead of cleaning my house. Suppose that California Labour Laws require that I pay you up-front and I know I have no chance of getting my money or stuff back.
Say your preference order is “Steal” > “Do the job” > “Don’t get hired”.
This is a racist dog-whistle which attacks ESL housekeepers; this sort of meme’s frustratingly common on the West Coast.
The third option is to exit the market. The underlying issue is that, regardless of how much effort one puts into cultivating non-criminal customers, VPN services will appear to attract criminals, and the optics alone is sufficient to dramatically limit and shape the pool of customers. Source: I operated some least-authority encrypted storage (I provably can’t read private customer data) from like 2016-2019 and my peer group thought I was facilitating criminals. Ironically, I left that peer group because they wouldn’t stop platforming Bitcoiners.
Small autosneer from Imgur. The warmup features images which I’ll alt-text here. The first image is a screenshot of a purchase of 128GB of DDR5 RAM, branded Crucial Pro, as 2 64GB sticks, priced at $305.95. The second image is a screenshot of the same 128GB of RAM, with the same model number, priced at $1800. The third image is a standard desktop personal-computer chassis, with side panels removed, revealing two gamer GPUs balanced above an ATX motherboard, all connected by a morass of cabling which has bulged out through the top and spilled over the edges, cooled by a large motherboard fan held on by plastic zip-ties. The post’s title:
In response to the top comment, which desires the bursting of the bubble:
And finally the punchline, in response to somebody pointing out that maybe there’s no good reason to buy top-of-the-line brand-new memory sticks: