Interview with gwern(dwarkeshpatel.com) |
Interview with gwern(dwarkeshpatel.com) |
> You have one Steve Jobs-type at the helm, and then maybe a whole pyramid of AIs out there executing it and bringing him new proposals
Very interesting in a short story (or a side quest in Cyberpunk 2077 - yeah that one). Not so much for a description of our future.
These AI predictions never, ever seem to factor in how actual humans will determine what AI-generated media is successful in replacing human-ones, or if it will even be successful at all. It is all very theoretical and to me, shows a fundamental flaw in this style of "sit in a room reading papers/books and make supposedly rational conclusions about the future of the world."
A good example is: today, right now, it is a negative thing for your project to be known as AI-generated. The window of time when it was trendy and cool has largely passed. Having an obviously AI-generated header image on your blog post was cool two years ago, but now it is passé and marks you as behind the trends.
And so for the prediction that everything get swept up by an ultra-intelligent AI that subsequently replaces human-made creations, essays, writings, videos, etc., I am doubtful. Just because it will have the ability to do so doesn't mean that it will be done, or that anyone is going to care.
It seems vastly more likely to me that we'll end up with a solid way of verifying humanity – and thus an economy of attention still focused on real people – and a graveyard of AI-generated junk that no one interacts with at all.
the argument against AI taking over is we organize around symbols and narratives and are hypersensitive to waning or inferior memes, thereofre AI would need to reinvent itself as "not-AI" every time so we don't learn to categorize it as slop.
I might agree, but if there were an analogy in music, some limited variations are dominant for decades, and there are precedents where you can generate dominant memes from slop that entrains millions of minds for entire lifetimes. Pop stars are slop from an industry machine that is indistinguishable from AI, and as evidence, current AI can simulate their entire catalogs of meaning. the TV Tropes website even identifies all the elements of cultural slop people should be immune to, but there are still millions of people walking around living out characters and narratives they received from pop-slop.
there will absolutely be a long tail of people whose ontology is shaped by AI slop, just like there is a long tail of people whose ontology is shaped by music, tv, and movies today. that's as close to being swept up in an AI simulation as anything, and perhaps a lot more subtle. or maybe we'll just shake it off.
But even if a future AI becomes like this, that doesn't prevent independent writers (like gwern) from still having a unique, non-assimilated voice where they write original content. The arguments tend to be "AI will eat everything, therefore get your writing out there now" and not "this will be a big thing, but not everything."
The pop industry is a:
- machine
- which takes authentic human meaning
- and produces essentially a stochastic echo of it ("slop")
- in an optimization algorithm
- to predict the next most profitable song (the song that is most "likely")
So, this sounds an awful lot like something else that's very in vogue right now. Only it was invented in 1950 or 1960, not in 2017.
Humanity is pinning its future on the thought that we will hit intractable information-theoretic limitations which provide some sort of diminishing returns on performance before a hard takeoff, but the idea that the currently demonstrated methods are high up on some sigmoid curve does not seem at this point credible. AI models are dramatically higher performance this year than last year, and were dramatically better last year than the year before, and will probably continue to get better for the next few years.
That's sufficient to dramatically change a lot of social & economic processes, for better and for worse.
Currently the state-of-the-art is propped up with speculative investments, if those speculations turn out to be wrong enough, or social/economic changes force the capital to get allocated somewhere else, then there could be a significant period of time where access to it goes away for most of us.
We can already see small examples of this from the major model providers. They launch a mind-blowing model, get great benchmarks and press, and then either throttle access or diminish quality to control costs / resources (like Claude Sonnet 3.5 pretty quickly shifted to short, terse responses). Access to SOTA is very resource-constrained and there are a lot of scenarios I can imagine where that could get worse, not better.
Even "Today, the state of the art in is the worst it will ever be" in cryptography isn't always true, like post-spectre/meltdown. You could argue that security improved but perf definitely did not.
But that isn’t the claim being made, which is that some sort of AI god is being constructed which will develop entirely without the influence of how real human beings actually act. This to me is basically just sci-fi, and it’s frankly kind of embarrassing that it’s taken so seriously.
Yeah, but, better at _what_?
Cars are dramatically faster today than 100 years ago. But they still can't fly.
Similarly, LLMs performing better on synthetic benchmarks does not demonstrate that they will eventually become superintelligent beings that will replace humanity.
If you want to actually measure that, then these benchmarks need to start asking questions that demonstrate superintelligence: "Here is a corpus of all current research on nuclear physics, now engineer a hydrogen bomb." My guess is, we will not see much progress.
This is true only because publicly-accessible models have been severely nerfed (out of sheer panic, one assumes), making their output immediately recognizable and instantly clichéd.
Dall-E 2, for instance, was much better when it first came out, compared to the current incarnation that has obviously been tweaked to discourage generating anything that resembles contemporary artists' output, and to render everything else in annoying telltale shades of orange and blue.
Eventually better models will appear, or be leaked, and then you won't be able to tell if a given image was generated by AI or not.
If, in the future, there is a way to validate humanity (as I mentioned in my comment), then any real writers will likely use it.
Anyone that doesn't validate their humanity will be assumed to be an AI. The reaction to this may or may not be negative, but the broader point is that in this scenario, the AI won't be eating all human creations.
I mean, you'd probably get more of a vote using generative AI to spam stuff that aligns with your opinions or moving to Kenya to do low wage RHLF stuff...
Still, I like a lot of his writing. Especially the weird and niche stuff that most people don’t even stop to think about. And thanks to Gwern’s essay on the sunk costs fallacy, I ended up not getting a tattoo that I had changed my mind about. I almost got it because I had paid a deposit, but I genuinely decided I hated the idea of what I was going to get… and almost got it, but the week before I went to get the tattoo, I read that essay, and decided if small children and animals don’t fall victim to sunk costs, then neither should I! Literally - Gwern saved the skin on my back with his writing. Haha.
The voice was uncanny. Simply hard to listen to, despite being realistic. I mean precisely that: it is cognitively difficult to string together meaning from that voice. (I am adjacent to the field of audio production and frequently deal with human- and machine-produced audio. The problem this podcast has with this voice is not unique.) The tonality and meaning do not support each other (this will change as children grow up with these random-tonality voices).
The conversation is excessively verbose. Oftentimes a dearth of reason gets masked by a wide vocabulary. For some audience members I expect the effort to understand the words distracts from the relationship between the words (ie, the meaning), and so it just comes across as a mashup of smart-sounding words, and the host, guest, and show gets lauded for being so intelligent. Cut through the vocabulary and occasional subtle tsks and pshaws and “I-know-more-than-I-am-saying” and you uncover a lot of banter that just does not make good sense: it is not quite correct, or not complete in its reasoning. This unreasoned conversation is fine in its own right (after all, this is how most conversation unfolds, a series of partially reasoned stabs that might lead to something meaningful), but the masking with exotic vocabulary and style is misleading and unkind. Some of these “smart-sounding” snippets are actually just dressed up dumb snippets.
Oh, how humbling and modest of him!
The voice was uncanny. Simply hard to listen to,
despite being realistic. I mean precisely that: it
is cognitively difficult to string together meaning
from that voice.
In order to protect Gwern's anonymity, I proposed
interviewing him in person, and having my friend Chris
Painter voice over his words after. This amused him
enough that he agreed.
But it did not sound fake to me, and the linked source indicates that it's a human.
Perhaps it sounds uncanny to you because it's a human reading a transcript of a conversation.... and attempting to make it sound conversational, as if he's not reading a transcript?
In addition to this, there are Lex Fridman's series of interviews with various key people from Anthropic [0], and a long discussion between Stephen Wolfram and Eliezer Yudkowsky on the theme of AI risk [1].
No! What’s wrong with Bitcoin is that it’s ugly. … It’s ugly to make your network’s security depend solely on having more brute-force computing power than your opponents, ugly to need now and in perpetuity at least half the processing power just to avoid double-spending … It’s ugly to have a hash tree that just keeps growing … It’s ugly to have a system which can’t be used offline without proxies and workarounds … It’s ugly to have a system that has to track all transactions, publicly … And even if the money supply has to be fixed (a bizarre choice and more questionable than the irreversibility of transactions), what’s with that arbitrary-looking 21 million bitcoin limit? Couldn’t it have been a rounder number or at least a power of 2? (Not that the bitcoin mining is much better, as it’s a massive give-away to early adopters. Coase’s theorem may claim it doesn’t matter how bitcoins are allocated in the long run, but such a blatant bribe to early adopters rubs against the grain. Again, ugly and inelegant.) Bitcoins can simply disappear if you send them to an invalid address. And so on.
First: actual visionary CEOs are a niche of a niche. Second: that is not how most companies work. The existence of the workforce is as important as what the company produces Third: who will buy or rent those services or products in a society where the most common economy driver (salaried work) is suddenly wiped out?
I am really bothered by these systematic thinkers whose main assumption is that the system can just be changed and morphed willy nilly as if you could completely disregard all of the societal implications.
We are surrounded by “thinkers” who are actually just glorified siloed-thinking engineers high on their own supply.
The link in this paragraph goes to a post on gwern website. This post contains various links, both internal and external. But I still failed to find one that supports claims about Newton's views on "progress".
> This offers a little twist on the “Singularity” idea: apparently people have always been able to see progress as rapid in the right time periods, and they are not wrong to! We would not be too impressed at several centuries with merely some shipbuilding improvements or a long philosophy poem written in Latin, and we are only modestly impressed by needles or printing presses.
We absolutely _are_ impressed. The concept of "rapid progress" is relative. There was rapid progress then, and there is even more rapid progress now. There is no contradiction.
Anyway, I have no idea how this interview got that many upvotes. I just wasted my time.
https://news.ycombinator.com/item?id=21559620
Again I'm sorry for the negativity, but already at the time Gwern was held up by a certain, large, section of the community as an important influencer in AI. For me that's just a great example of how basically the vast majority of AI influencers (who vie for influence on social media, rather than research) are basically clueless about AI and CS and only have second-hand knowledge, which I guess they're good at organising and popularising, but not more than that. It's easy to be a cheer leader for the mainstream view on AI. The hard part is finding, and following, unique directions.
With apologies again for the negative slant of the comment.
In my country, when TV series are interviewing anonymous people they use specific visual language - pixellated face, or facing away from the camera, or face clad in shadow.
Having an actor voice the words is normal. But having an actor showing the anonymous person's face is an... unusual choice.
I know it's probably a "me problem" ;)
Downsides of anonymity: no free heroin
Something I've noticed in spending time online is that there's a "core group" of a few dozen people who seem to turn up everywhere there are interesting discussions. Gwern (who also posts here) is probably at the top of that list.
> In Internet culture, the 1% rule is a general rule of thumb pertaining to participation in an Internet community, stating that only 1% of the users of a website actively create new content, while the other 99% of the participants only lurk. Variants include the 1–9–90 rule (sometimes 90–9–1 principle or the 89:10:1 ratio),[1] which states that in a collaborative website such as a wiki, 90% of the participants of a community only consume content, 9% of the participants change or update content, and 1% of the participants add content.
I might be reading to much into the "a" (rather than "the"), but to be clear: there is a transcript at the bottom.
>I live in the middle of nowhere. I don't travel much, or eat out, or have health insurance, or anything like that. I cook my own food. I use a free gym. There was this time when the floor of my bedroom began collapsing. It was so old that the humidity had decayed the wood. We just got a bunch of scrap wood and a joist and propped it up. If it lets in some bugs, oh well! I live like a grad student, but with better ramen. I don't mind it much since I spend all my time reading anyway.
Not sure what to think of that. On one hand, it's so impressive that gwern cares only about the intellectual pursuit. On the other hand, it's sad that society does not reward it as much as excel sheet work.
Dwarkesh has 18 splits. https://www.dwarkeshpatel.com/i/151435243/timestamps
I got 171. So roughly 9 context discussions in one time stamp.
https://gpt3experiments.substack.com/p/a-new-chunking-approa...
Then when I saw the frankly very creepy and offputting image and voice, thinking he'd been anonymised through some AI software, thought, oh no, this kind of thing isn't going to become normal is it.
Then - plot twist - I scroll down to read the description and see that that voice is an actual human voiceover! I don't know if that makes it more or less creepy. Probably more. What a strange timeline.
Gwern is an effective altruist and his influence is largely limited to that community. It would be an exaggeration to claim that he influenced the mainstream of AI and ML researchers -- certainly Hinton, LeCun, Ng, Bengio didn't need him to do their work.
He influences the AI safety crowd, who have ironically been trying to build AGI to test their AI safety ideas. Those people are largely concentrated at Anthropic now, since the purge at OpenAI. They are poorly represented at major corporate AI labs, and cluster around places like Oxford and Cal. The EAs' safety concerns are a major reason why Anthropic has moved so much slower than its competitors, and why Dario is having trouble raising the billions he needs to keep going, despite his media blitz. They will get to AGI last, despite trying to be the good guys who are first to invent god in a bottle.
By the same token, Dwarkesh is either EA or EA adjacent. His main advertiser for this episode is Jane Street, the former employer of the world's most notorious EA, Sam Bankman-Fried as well as Caroline Ellison. Dwarkesh previously platformed his friend Leopold Aschenbrenner, who spent a year at OAI before he wrote the scare piece "Situation Report" made the rounds. Leopold is also semi-technical at best. A wordcel who gravitated to the AI narrative, which could describe many EAs.
People outside of AI and ML, please put Dwarkesh in context. He is a partisan and largely non-technical. The way he interfaces with AI is in fantasizing about how it will destroy us all, just as he and Gwern do in this interview.
It's sad to see people who are obviously above average intelligent waste so much time on this.
He says he has collaborators under the "Gwern" name now, but the main guy is the main guy and it's unlikely he could hide it.
How many citations for "Branwen 2018" are on the ArXiv now?
Neither mpv nor my browser can play the interview.
Edit: The YouTube link from an earlier submisison works: <https://www.youtube.com/watch?v=a42key59cZQ>.
Sad commentary when a YouTube link affords greater accessibility (I'm listening now via mpv) than a podcast URL.
Sigh.
Thing is: Are you absolutely sure that notion of human biodiversity is wrong? IQ is heritable, as height is heritable. You'll grant that there are populations that differ in their genetic potential for height -- e.g. Dalmatians vs. Pygmies -- so how is it that you dismiss out of hand the notion that there might be population-wide differences in the genetic potential for intelligence?
I can hear it now: "But IQ is not intelligence!" I agree to a point, but IQ -- and, strangely, verbal IQ in particular -- maps very neatly to one's potential for achievement in all scientific and technological fields.
The Truth is a jealous goddess: If you devote yourself to her, you must do so entirely, and take the bad along with the good. You don't get to decide what's out of bounds; no field of inquiry should be off-limits.
I'm not gonna say that people there don't think that hbd is real, but it's not an everyday discussion topic. Mostly because it's kind of boring.
(Do spend five minutes on #lesswrong! We don't bite! (Generally! If you come in like "I heard this was the HBD channel", there may be biting, er, banning.))
"In order to protect Gwern's anonymity, I proposed interviewing him in person, and having my friend Chris Painter voice over his words after. This amused him enough that he agreed."
Gwern, how did you end up with free gym? Interested for myself.
Btw, the article gets it right and starts with 'Gwern is a pseudonymous researcher and writer.'
Headlines are seldom decided by the writers.
This leads me to believe that the content and the subsequent posts are for self promotion purposes only.
The worst I can say is that I find his predictions around AI (i.e. the scaling laws) to be concerning.
edit: having now read the linked interview, I can provide a clearly non-misanthropic quote, in response to the interviewer asking gwern what kind of role he hopes to play in people's lives:
I would like people to go away having not just been entertained or gotten some useful information, but be better people, in however slight a sense. To have an aspiration that web pages could be better, that the Internet could be better: “You too could go out and read stuff! You too could have your thoughts and compile your thoughts into essays, too! You could do all this!”But I do know he created an enormous dataset of anime images used to train machine learning and generative AI models [1]. Hosting large datasets is moderately expensive - and it's full of NSFW stuff, so he's probably not having his employer or his college host it. Easy for someone on a six-figure salary, difficult for a person on $12k/year.
Also, I thought these lesswrong folks were all about "effective altruism" and "earning to give" and that stuff.
I don't like that now people might pigeonhole him a bit by thinking about his effective frugality but I do hope he gets a ton of donations (either directly or via patreon.com/gwern ) to make up for it.
EDIT: grammar
> Gwern was the first patient to successfully complete a medical transition to the gender he was originally assigned at birth... his older brother died of a Nuvigil overdose in 2001... his (rather tasteful) neck tattoo of the modafinil molecule
The only concrete things we know about gwern are that he's a world-renowned breeder of Maine Coons and that he is the sole known survivor of a transverse cerebral bifurcation.
He does have a neck tattoo, but it's actually a QR code containing the minimal weights to label MNIST at 99% accuracy.
I haven't decided whether I agree with it, but I can see the thought behind it: the more mechanical work will be automated, but long-term direction setting will require more of a thoughtful hand.
That being said, in a full-automation economy like this, I imagine "AI companies" will behave very differently to human companies: they can react instantly to events, so that a change in direction can be affected in hours or days, not months or years.
Where is the data showing that more jobs get destroyed than created by technological disruption?
Without saying anything regarding the arguments for or against AI, I will address this one sentence. This quote is an example of an appeal to hypocrisy in history fallacy, a form of the tu quoque fallacy. Just because someone criticizes X and you compare it to something else (Y) from another time does not mean that the criticism of X is false. There is survivorship bias as well because we now have cars, but in reality, you could've said this same criticism against some other thing that failed, but you don't, because, well, it failed and thus we don't remember it anymore.
The core flaw in this reasoning is that just because people were wrong about one technology in the past doesn't mean current critics are wrong about a different technology now. Each technology needs to be evaluated on its own merits and risks. It's actually a form of dismissing criticism without engaging with its substance. Valid concerns about X should be evaluated based on current evidence and reasoning, not on how people historically reacted to Y or any other technology.
I'm not sure that it's acrually correct: I don't think we'll actually see "AI" actually replace work in general as a concept. Unless it can quite literally do everything and anything, there will always be something that people can do to auction their time and/or health to acquire some token of social value. It might taken generations to settle out who is the farrier who had their industry annihilated and who is the programmer who had it created. But as long as there's scarcity and ambition in the world, there'll be something there, whether it's "good work" or demeaning toil under the bootheel of a fabulously wealthy cadre of AI mill owners. And there will be scarcity as long as there's a speed of light.
Even if I'm wrong and there isn't, that's why it's called the singularity. There's no way to "see" across such an event in order to make predictions. We could equally all be in permanent infinite bliss, be tortured playthings of a mad God, extinct, or transmuted into virtually immortal energy beings or anything in between.
You might as well ask the dinosaurs whether they thought the ultimate result of the meteor would be pumpkin spice latte or an ASML machine for all the sense it makes.
Anyone claiming to be worrying over what happens after a hypothetical singularity is either engaging in intellectual self-gratification, posing or selling something somehow.
All you really need is a government or society that isn't conducive to technological development, either because they persecute it or because they just don't do anything to protect and encourage it (e.g. no patent system or enforceable trade secrets).
Even today, what we see is that technological progress isn't evenly distributed. Most of it comes out of the USA at the moment, a bit from Europe and China. In the past there's usually been one or two places that were clearly ahead and driving things forward, and it moves around over time.
The other thing that inspires the idea of a permanent medieval society is archaeological narratives about ancient Egypt. If you believe their chronologies (which you may not), then Egyptian society was frozen in time for thousands of years with little or no change in any respect. Not linguistic, not religious, not technological. This is unthinkable today but is what academics would have us believe really happened not so long ago.
Not discovering sources of cheap energy and other raw inputs. If you look carefully at history, every rapid period of growth was preceded by a discovery or conquest of cheap energy and resources. You need excess to grow towards the next equilibrium.
https://en.wikipedia.org/wiki/House_of_Wisdom#Destruction_by...
https://en.wikipedia.org/wiki/Library_of_Alexandria#Burning_...
https://en.wikipedia.org/wiki/Nalanda_mahavihara#Destruction...
https://en.wikipedia.org/wiki/Imperial_Library_of_Constantin...
https://en.wikipedia.org/wiki/List_of_destroyed_libraries#Hu...
Could you explain further what part of https://gwern.net/newton you thought didn't support my description of Newton's view?
I thought the large second blockquote in https://gwern.net/newton#excerpts , which very prominently links to https://www.newtonproject.ox.ac.uk/view/texts/normalized/THE... , fully justified my claims, which are taken directly from Newton's statements to his son-in-law, and also closely parallel other historical statements, like Lucretius, which I also present with clear references and specific blockquotes.
I'm a little mystified that you could describe any of this as not supporting it at all, and I'm wondering if you are looking at the wrong page or something?
Irrespective of the historical accuracy of the quote I've always felt this way in some form, having personally lived through the transition from a world where it felt like you didn't have to have an opinion on everything to one dominated by the ubiquitous presence of the Internet. Although not so much because I believe an advanced human civilization has destroyed itself in our current timeline, but because the presence of so many life-changing breakthroughs in such a short period of time to me indicates a unceasing march towards a Great Filter.
Then there isn't a downvote option if it proves poor.
This is a bit stark: there are many great knowledgeable engineers and scientists who would not get your point about a^nb^n. It's impossible to know 100% of of such a wide area as "AI and CS".
I think, engineers, yes, especially those who don't have a background in academic CS. But scientists, no, I don't think so. I don't think it's possible to be a computer scientist without knowing the difference between a regular and a super-regular language. As to knowing that a^nb^n specifically is context-free, as I suggest in the sibling comment, computer scientists who are also AI specialists would recognise a^nb^n immediately, as they would Dyck languages and Reber grammars, because those are standard tests of learnability used to demonstrate various principles, from the good old days of purely symbolic AI, to the brave new world of modern deep learning.
For example, I learned about Reber grammars for the first time when I was trying to understand LSTMs, when they were all the hype in Deep Learning, at the time I was doing my MSc in 2014. Online tutorials on coding LSTMs used Reber grammars as the dataset (because, as with other formal grammars it's easy to generate tons of strings from them and that's awfully convenient for big data approaches).
Btw that's really the difference between a computer scientist and a computer engineer: the scientist knows the theory. That's what they do to you in CS school, they drill that stuff in your head with extreme prejudice; at least the good schools do. I see this with my partner who is 10 times a better engineer than me and yet hasn't got a clue what all this Chomsky hierarhcy stuff is. But then, my partner is not trying to be an AI influencer.
totally agree that you can be a great engineer and not be familiar with it, but seems weird for an expert in the field to confidently make wrong statements about this.
Transformers can easily intellectually understand a^nb^n, even though they couldn't recognize whether an arbitrarily long string is a member of the language -- a restriction humans share!, since eventually a human, too, would lose track of the count, for a long enough string.
>> Regarding your linked comment, my takeaway is that the very theoretical task of being able to recognize an infinite language isn't very relevent to the non-formal, intuitive idea of "intelligence"
That depends on who you ask. My view is that automata are relevant to computation and that's why we study them in computer science. If we were biologists, we would study beetles. The question is whether computation , as we understand it on the basis of computer science, has anything to do with intelligence. I think it does, but that it's not the whole shebang. There is a long debate on that in AI and the cognitive sciences and the jury is still out, despite what many of the people working on LLMs seem to believe.
Yes, LLMs are bad at this. A similar example: SAT solvers can't solve the pigeonhole problem without getting into a loop
It is an exceptional case that requires "metathinking" maybe, rather than a showstopper issue
(can't seem to be able to write the grammar name, the original comment from the discussion had it)
Appreciate the diversity in the effort, but engineering is making things people can use without having to know it all. Far more interesting endeavor than being a human Google search engine.
And there's a limit to what you need to look up in a book. The limit moves further up the more you work with a certain kind of tool or study a certain kind of knowledge. I have to look up trigonometry every single time I need it because I only use it sparingly. I don't need to look up SLD-Resolution, which is my main subject. How much would Feynman need to look up when debating physics?
So when someone like Feynman talks about physics, you listen carefully because you know they know their shit and a certain kind of nerd deeply appreciates deep knowledge. When someone elbows themselves in the limelight and demands everyone treats them as an expert, but they don't know the basics, what do you conclude? I conclude that they're pretending to know a bunch of stuff they don't know.
________________
[1] ... some do. But they're students so it's OK, they're just excited to have learned so much and don't yet know how much they don't. You explain the mistake, point them to the book, and move on.
In the thread you linked, Gwern says in response to someone else that NNs excel at many complex real-world tasks even if there are some tasks where they fail but humans (or other models) succeed. You try to counter that by bringing up an example for the latter type of task? And then try to argue that this proves Gwern wrong?
Whether they said "regular grammar" or "context-free grammar" doesn't even matter, the meaning of their message is still the exact same.
Edit: sorry, I read "finite" as "infinite" :0 But n can be infinite and a^nb^n is still regular, and also context free. To be clear, the Chomskky Hierarchy of formal languages goes like this:
Finite ⊆ Regular ⊆ Context-Free ⊆ Context-Sensitive ⊆ Recursively Enumerable
That's because formal languages are identified with the automata that accept them and when an automaton accepts e.g. the Recursively Enumerable languages, then it also accepts the context-sensitive languages, and so on all the way down to the finite languages. One way to think of this is that an automaton is "powerful enough" to recognise the set of strings that make up a language.
Specifically, you can construct a finite automata to represent it.
The goal of influencers is to influence the segment of a crowd who cares about influencers. Meaning retards and manchildren looking for an external source to form consensus around.
He's been building software for 10 years longer than I've been alive, hopefully in a few decades I'll have gained the same breadth of technical perspective he's got.
I don't know what causes such intellectual cliques to form, perhaps it's a result of an intersection of raw intellectual power and social dynamics.
> How do you sustain yourself while writing full time?
> Gwern
> Patreon and savings. I have a Patreon which does around $900-$1000/month, and then I cover the rest with my savings. I got lucky with having some early Bitcoins and made enough to write for a long time, but not forever. So I try to spend as little as possible to make it last.
Then Dwarkesh just gets stuck on this $1k/month thing when Gwern right out of gate said that savings are being used.
Who knows how much of the savings are being used or how big of a profit he got from BTC.
He's living in a place where the floor collapsed and eats (good) ramen. If it's 12k or 20k I'm not sure it makes a meaningful difference to the narrative.
https://en.wikipedia.org/wiki/Poverty_in_the_United_States
"Among individuals living alone: 19.1% lived in poverty."
Poverty line (max yearly income) for single households: $14,580
Especially if you are doing it voluntarily, 1k a month can provide you more then enough for a comfortable life in many part of the country. More so if you can avoid car ownership/insurance and health insurance (Which gwern seems to do).
A room is only a prison cell when you're not allowed to leave.
It's a reasonable tradeoff for some circumstances.
Maybe it works for maths, physics and such, and of course it's ok to philosophize, but I think those "ivory tower" thinkers sometimes lack a certain connection to reality
Unless by reality you mean YOUR slice of the world, a sheltered place of its own.
In what sense? Claude is way better than any other LLM available, in the sense of providing useful/correct/intelligent output.
Gwern's probably not a "wordcel" either, he can program, right? I've never seen any of his publications though.
It's called Situational Awareness too, not "Situation Report", and Yudkowsky said he didn't like it. Not that Yudkowsky is an EA either.
I think the situation is more complex than you think it is, or at least more complex than you're pretending.
Edit: oh, you're saying Gwern is an EA too? Do you have a source for that?
I didn't claim Yudkowsky was an EA. He's not. He's a rationalist who originated many thoughts about AI doom, which have influenced EAs.
Look, Gwern and Dwarkesh and Yud are all AI doomers, which is where the rationalists and the EAs overlap. Gwern is all over LessWrong, which kind of springs out of the rationalist community and has all sorts of EAs there now. But of course, the first rule of EA is the first rule of Fight Club. That's why Dario consistently denies he's an EA, even though his chief of staff was formerly running Sam Bankman-Fried's philanthropic grants.
The weird thing in these conversations between guys like Dwarkesh and Gwern is how much they wallow in the masochism of imagining a world taken over by AGI. Yud has a similar attraction to pain, but he channels it into his silly BDSM novels.
https://www.amazon.com/Dark-Lords-Answer-Eliezer-Yudkowsky-e...
If that's how they wan't to spend their time, fine. Let's all just recognize it's an acquired taste, a self-indulgence, and the furthest thing from a rational act.
He does Haskell stuff (he mentions it on his website), but he’s even better at words than he is as a mid-tier (I guess) Haskell programmer.
Whether this counts as a “wordcel” is an exercise left to the reader.
Good luck.
Although the avatar tool is probably not SOTA, I thought the 3D model was a really cool way to deal with interviewing Gwern, I am quite enjoying the current video.
Any technology capable of distinguishing between machine- and human-created text or artwork will simply be placed in a feedback loop and used for training the next generation of models.
If this indeed turns out to be a fight, humans won't win it. So let's not bother with the fighting part.
The alternative is that everyone just accepts the internet as a place where you can’t tell if anyone is real. I don’t think that will be a happy state of affairs for anyone.
Even if this isn’t a “required” thing, it will become a desired one by consumers. Who would you rather follow - the verified Taylor Swift account, or the unverified one? Non-verified creators will be relegated to niches where it doesn’t matter as much if they’re real or not.
Then you might say - but a person can just use AI tools and paste them into their verified account. Which is fine - I’m not saying that people aren’t going to use these tools to create content. I’m saying that these tools aren’t going to eliminate the possibility of an individual person creating content on the Internet as a human being, and being verified as such.
AI maximalists are talking about breaking that pattern and having AI literally do the job and provide the service, cutting out the need for workers entirely. Services being provided entirely autonomously and calories being generated without human input in the two analogies.
I'm not convinced by that at all: if services can be fully automated, who are you going to sell ERP or accounting software to, say? What are people going to use as barter for those calories if their time and bodies are worthless?
But I can see why that is a saleable concept to those who consider the idea of needing to pay workers to be a cruel injustice. Though even if it works at all, which, as I said, I dont believe, the actual follow-on consequences of such a shift are impossible to make any sensible inferences about.
None match that thesis
Age of exploration was powered by finding new sources of slaves and materials in the East (india, asia, also eastern europe to an extent)
The mongol conquest itself was capturing vast sources of wealth (easier to take from others than build yourself)
Same with Rome. Each new conquest brought more slaves and natural resources to the empire. It used this to fuel more expansion.
If you only need a few terabytes, you can rent from Hetzner for more like <$20/month. Maybe <$10 at this point, if you are patient. And you can use the server for other things too, since it's a fixed cost, like your website or encrypted backups or other files you host. I spent a while figuring this out before I decided to do Danbooru20xx, to make sure it was very cheap. Given how my website has grown in size, and how absurdly exorbitant cloud bandwidth and object/hard drive space is, the Danbooru dataset was practically free at the margin!
> Also, I thought these lesswrong folks were all about "effective altruism" and "earning to give" and that stuff.
Some people are, some people aren't. I'm not. (I have intellectual sympathy with EA, but not emotional.)
Edit: dude, come on. That's no way to have a debate. Other times I'm the one who gets all the downvotes. You gotta soldier on through it and say your thing anyway. Robust criticism is great but being prissy about downvotes just makes HN downvote you more.
“Debating physics” in academia is little more than reciting preferred symbolic logic. The map is not the terrain; to borrow from Feynman again, who says in his lectures it’s not good to get hung up on such stuff. So you’re just making this about rhetorical competition, boring gamesmanship. Which like I said, isn’t the point of engineering. Dialectics don’t ship, they keep professors employed though.
Whichever way it’s written, physics still seems to work. So the value of STEM is more about the application than the knowing.
Turning that into an agent which builds its own hydrogen bomb using what amount to seized resources and to do it covertly at a pace that is faster than human agencies notice is a different sort of thing, but the middleware to do that sort of agent directed project is rapidly developing as well, and there is strong economic incentive for self-interested actors to pursue it. For a very brief moment in time, a huge amount of shareholder value will be created, and then suddenly destroyed.
A large-scale nuclear exchange is no longer the worst case scenario, in point of fact.
Assuming we don't hit those information-theoretic barriers, and we don't develop a host of new safeguards which nobody at the present time seems interested in developing.
ok buddy
https://en.wikipedia.org/wiki/Thermonuclear_weapon
It's a mechanistic process featuring well-tread factual, verbose discourse. Scientists reasoning about facts and presenting what they found. "Tell me a joke about elephant mating habits in the form of a rap song" is a dramatically more complex task of synthesis.
> https://extension.missouri.edu/publications/g2910
Heritability is simply a measure of how much of the variation in a trait, like height or IQ, is due to genetic factors rather than environmental influences in a given population. Could be feedlot steers, could be broiler chickens, could be humans. In humans, traits like height are very highly heritable, at ~0.8. Certain others, like eye color, are heritable to ~0.98.
Differences in IQ between groups are apparently far more modest, but, however distasteful, it's still possible to speak of them, and it's possible to make statistical statements about them. My position is simply that, on the one side, it should be done in good faith -- and, on the other side, it shouldn't be seen as something heretical.
How do you know those things are true?
That stuff is still absolutely relevant, btw. Some DL people like to dismiss it as irrelevant but that's just because they lack the background to appreciate why it matters. Also: the arrogance of youth (hey I've already been a postdoc for a year, I'm ancient). Here's a recent paper on Neural Networks and the Chomsky Hierarchy that tests RNNs and Transformers on formal languages (I think it doesn't test on a^nb^n directly but tests similar a-b based CF languages):
https://arxiv.org/abs/2207.02098
And btw that's a good paper. Probably one of the most satisfying DL papers I've read in recent years. You know when you read a paper and you get this feeling of satiation, like "aaah, that hit the spot"? That's the kind of paper.
A transformer (with relative invariant positional embedding) has full context so can see the whole sequence. It just has to count and compare.
To convince yourself, construct the weights manually.
First layer :
zeros the character which are equal to the previous character.
Second layer :
Build a feature to detect and extract the position embedding of the first a. a second feature to detect and extract the position embedding of the last a, a third feature to detect and extract the position embedding of the first b, a fourth feature to detect and extract the position embedding of the last b,
Third layer :
on top that check whether (second feature - first feature) == (fourth feature - third feature).
The paper doesn't distinguish between what is the expressive capability of the model, and the finding the optimum of the model, aka the training procedure.
If you train by only showing example with varying n, there probably isn't inductive bias to make it converge naturally towards the optimal solution you can construct by hand. But you can probably train multiple formal languages simultaneously, to make the counting feature emerge from the data.
You can't deduce much from negative results in research beside it requiring more work.
They do. That's the whole point of the paper: you can set a bunch of weights manually like you suggest, but can you learn them instead; and how? See the Introduction. They make it very clear that they are investigating whether certain concepts can be learned by gradient descent, specifically. They point out that earlier work doesn't do that and that gradient descent is an obvious bit of bias that should affect the ability of different architectures to learn different concepts. Like I say, good work.
>> But you can probably train multiple formal languages simultaneously, to make the counting feature emerge from the data.
You could always try it out yourself, you know. Like I say that's the beauty of grammars: you can generate tons of synthetic data and go to town.
>> You can't deduce much from negative results in research beside it requiring more work.
I disagree. I'm a falsificationist. The only time we learn anything useful is when stuff fails.
The fact that similar, often more detailed assertions of the imminent disappearance of work has been a consistent trope since the beginning of the Industrial Revolution (as acknowledged in literally the next question in the interview, complete with an interestingly wrong example) and we've actually ended up with more jobs seems far more like a relevant counterargument than ad hominem tu quoque...
A floor collapsing and not bothering to replace it sounds more serious, sure, but that can mean a lot of different things in different circumstances. Imagine, for example, someone who's an expert in DIY but also a devoted procrastinator. That person could leave a roof in the state described for months or years, planning to eventually do it up, and I wouldn't consider anything terribly revelatory about the person's financial or mental status to have occurred.
https://paulgraham.com/ramenprofitable.html
This is separate from Japanese-style ramen, which you can also easily get at many restaurants.
The narrative is that he isn't living a life of luxury or even comfort.
He has priorities other than his comfort.
That's the narrative and from that viewpoint it isn't important if he's living on 12k a year or 20k a year because of his BTC investment.
But living on ramen + $12k/year + no healthcare + collapsing floor? That's sounding a lot like poverty to me.
- ramen =/= cheap pot noodle type things. Ramen restaurants in Japan attest to this. It'll depend on context what that actually means in this case.
- 12k/year =/= no money, numbers like that make no sense outside of context. It depends on how much of it is disposable. You live in a hut with no rent, you grow food, you've no kids, no mortgage, no car, etc, these all matter.
- no healthcare =/= bad health. How many of the growing numbers of people dying from heart-related diseases in their 50s and 60s had healthcare? Didn't do much for them, in the end.
- collapsing floor =/= bad hygiene or something else actually potentially dangerous, as I said above, the nuisance this causes or doesn't cause depends on lots of actual real factors, it's not some absolute thing. It just sounds wild to people not used to it
The physical principles remain regardless of how humans write them down.
I mean obviously it happened when I moved from industry to academia, but it's still the case there's a lot of overlap between the two areas, at least in CS and AI. In CS and AI the best engineers make the best scientists and vv. I think.
Btw, "gatekeeping" I think assumes that I somehow think of one category less than the other? Is that right? To be clear, I don't. I was responding to the use of both terms in the OP's comments with a personal reflection on the two categories.
The current wisdom is that by optimizing for multiple tasks simultaneously, it makes the energy landscape smoother. One task allow to discover features which can be used to solve other tasks.
Useful features that are used by many tasks can more easily emerge from the sea of useless features. If you don't have sufficiently many distinct tasks the signal doesn't get above the noise and is much harder to observe.
That the whole point of "Generalist" intelligence in the scaling hypothesis.
For problems where you can write a solution manually you can also help the training procedure by regularising your problem by adding the auxiliary task of predicting some custom feature. Alternatively you can "Generatively Pretrain" to obtain useful feature, replacing custom loss function by custom data.
The paper is a useful characterisation of the energy landscape of various formal tasks in isolation, but doesn't investigate the more general simpler problem that occur in practice.
This is my point, it is not known whether those are all in the same category. The survivorship bias part is that we only know if something were disruptive after the fact, because, well, they disrupted. Therefore, you cannot even compare them like that, never mind the fact that all disruptive technologies are not the same, so you shouldn't be comparing between them anyway.
(Of course, maybe you could argue that's a famous example in its training set and it's just regurgitating, but then you could try making modifications, asking other questions, etc, and the LLM would continue to respond sensibly. So to me it seems to understand...)
Or going back to the original Hofstadter article, "simple tests show that [machine translation is] a long way from real understanding"; I tried rerunning the first two of these simple tests today w/ Claude 3.5 Sonnet (new), and it absolutely nails them. So it seems to understand the text quite well.
Regarding computation and understanding: I just though it was interesting that you presented a true fact about the computational limitations of NNs, which could easily/naturally/temptingingly -- yet incorrectly (I think!) -- be extended into a statement about the limitations of understanding of NNs (whatever understanding means -- no technical definition that I know of, but still, it does mean something, right?).
Yes, well, that's the big confounder that has to be overcome by any claim of understanding (or reasoning etc) by LLMs, isn't it? They've seen so much stuff in training that it's very hard to know what they're simply reproducing from their corpus and what not. My opinion is that LLMs are statistical models of text and we can expect them to learn the surface statistical regularities of text in their corpus, which can be very powerful, but that's all. I don't see how they can learn "understanding" from text. The null hypothesis should be that they can't and, Sagan-like, we should expect to see extraordinary evidence before accepting they can. I do.
>> Regarding computation and understanding: I just though it was interesting that you presented a true fact about the computational limitations of NNs, which could easily/naturally/temptingingly -- yet incorrectly (I think!) -- be extended into a statement about the limitations of understanding of NNs (whatever understanding means -- no technical definition that I know of, but still, it does mean something, right?).
For humans it means something- because understanding is a property we assume humans have. Sometimes we use it metaphorically ("my program understands when the customer wants to change their pants") but in terms of computation... again I have no clue.
I generally have very few clues :)
There's also stuff like the Golden Gate Claude experiment and research @repligate shares on twitter, which again make me think understanding (as I conceive of it) is definitely there.
Now, are the conscious, feeling entities? That is a harder question to answer...
Edit: and technically you're describing what is more or less backprop learning, neural networks, by themselves, don't learn at all.
>> How do you do intelligence without computation though?
Beats me! Unlike everyone else in this space, it seems, I haven't got a clue how to do intelligence at all, with or without computation.
Edit: re infinite languages, I liked something Walid Saba (RIP) pointed out on Machine Learning Street Talk, that sure you can't generate infinite strings but if you have an infinite language every string accepted by the language has a uniform probability of one over infinity, so there's no way to learn the entire language by learning the distribution of strings within it. But e.g. the Python compiler must be able to recognise an infinite number of Python programs as valid (or reject those that aren't) because of the same reason, that it's impossible to predict which string is going to come out of a source generating strings in an infinite language. So you have to able to deal with infinite possibilities, with only finite resources.
Now, I think there's a problem with that. Assuming a language L has a finite alphabet, even if L is infinite (i.e. it includes an infinite number of strings) the subset of L where strings only go up to some length n is going to be finite. If that n is large enough that it is just beyond the computational resources of any system that has to recognise strings in L (like a compiler) then any system that can recognise, or generate, all strings in L up to n length, will be, for all intents and purposes, complete with respect to L, up to n etc. In plain English, the Python compiler doesn't need to be able to deal with Python programs of infinite length, so it doesn't need to deal with an infinite number of Python programs.
Same for natural language. The informal proof of the infinity of natural language I know of is based on the observation that we can embed an arbitrary number of sentences in other sentences: "Mary, whom we met in the summer, in Fred's house, when we went there with George... " etc. But, in practice, that ability too will be limited by time and human linguistic resources, so not even the human linguistic ability really-really needs to be able to deal with an infinite number of strings.
That's assuming that natural language has a finite alphabet, or I guess lexicon is the right word. That may or may not be the case: we seem to be able to come up with new rods all the time. Anyway some of this may explain why LLMs can still convincingly reproduce the structure of natural language without having to train on infinite examples.
And you can't get away with the standard tricky tricks that people use to say it isn't easy, logical induction exists.
That aside, we're getting into semantics. Whether you call it "genetic determinism" or "heritability," we're talking about durable group differences in genetically-mediated traits. And that is what people may find distasteful or even heretical.
The semantics matter, because the evidence supporting HBD positions is stated in terms of the technical definition of heritability.
While I've got you, can I ask that you stop evoking "heresy" and "distaste" in this thread? I believe I'm making simple, objective points, not summoning opprobrium on your position.
I agree that 12k goes a long way when you're a subsistence farmer, living alone in a collapsing hut in the middle of nowhere, with no health insurance.
Nonetheless, that's sounding a lot like poverty to me.
"no health insurance" is the only thing in your list there that isn't hyperbole or a twisting of what was said. But anyway, again, no-one is arguing about whether 12,000/year "officially" is or is not below the poverty line, except you, with yourself.
Did you read the part about him having savings in bitcoin that wouldn't sustain him forever, but nonetheless for many, many years? If the bitcoin was worth 500,000, for example, would you then say "oh, that's a totally acceptable way to live now!", or would you still be belittling their life choices, insinuating there was something bad or wrong with it?
I'm just saying what's described in the article sounds a great deal like poverty.
> Yeah, it is mostly just money stopping me at this point. I probably should bite the bullet and move anyway. But I'm a miser at heart and I hate thinking of how many months of writing runway I'd have to give up for each month in San Francisco.
> If someone wanted to give me, I don’t know, $50–100K/year to move to SF and continue writing full-time like I do now, I'd take it in a heartbeat.
It sounds a little like you are arguing without any reference to the actual evidence here.
But traits like IQ, height, and eye color are both (A) highly heritable and (B) substantially shaped by genetic factors. In casual online discourse, I believe that (B) is usually taken for granted, so it's glossed over, and when people say that any given trait is "heritable" they're also assuming that (B) is true for the trait. At least, I am guilty of that lapse.
And I take your point about language.
There is growing evidence that group IQ heritability isn't evidence of genetic causation.
First sentence of the conclusion: "Genetic association studies have confirmed a century of quantitative genetic research showing that inherited DNA differences are responsible for substantial individual differences in intelligence test scores."
Related: https://www.pnas.org/doi/full/10.1073/pnas.1408777111
The trouble is that, like height, IQ is governed by a vast network of "genes of small effect," so a comprehensive view has proven difficult to nail down. Progress is apparently being made, though slowly.
> There is growing evidence that group IQ heritability isn't evidence of genetic causation.
What evidence do you speak of?
And you should see the criticism I get by other academics when I try to publish my papers and they decide I'm not even wrong. And that kind of criticism has teeth: my papers don't get published.
>> Please be aware that your criticism has teeth too, you just don't feel the bite of them.
Well, maybe it does. I don't know if that can be avoided. I think most people don't take criticism well. I've learned for example that there are some conversations I can't have with certain members of my extended family because they're not used to being challenged about things they don't know and they react angrily. I'm specifically remember a conversation where I was trying to explain the concept of latent hypoxia and ascent blackout [1] (I free dive recreationally) to an older family member who is an experienced scuba diver, and they not only didn't believe me, they called me an ignoramus. Because I told them something they didn't know about. Eh well.
_____________
[1] It can happen that while you're diving deep, the pressure of the water keeps the pressure of oxygen in your blood sufficient that you don't pass out, but then when you start coming up, the pressure drops and the oxygen in your blood thins out so much that you pass out. In my lay terms. My relative didn't believe that the water pressure affects the pressure of the air in your vessels. I absolutely can feel that when I'm diving- the deeper I go the easier it gets to hold my breath and it's so noticeable because it's so counter-intuitive. My relative wouldn't have experienced that during scuba diving (since they breathe pressurised air, I think) and maybe it helps he's a smoker. Anyway I never managed to convince him.
As I never managed to convince him that we eat urchins' genitals, not their eggs. After a certain point I stopped trying to convince him of anything. I mean I felt like a know-it-all anyway, even if I knew what I was talking about.
My formative experience as a PhD student was when a senior colleague attacked my work. That was after I asked for his feedback for a paper I was writing where I showed that my system beat his system. He didn't deal with it well, sent me a furiously critical response (with obvious misunderstandings of my work) and then proceeded to tell my PhD advisor and everyone else in a conference we were attending that my work is premature and shouldn't be submitted. My advisor, trusting his ex-student (him) more than his brand new one (me), agreed and suggested I should sit on the paper a bit longer.
Later on the same colleague attacked my system again, but this time he gave me a concrete reason why: he gave me an example of a task that my system could not complete (learn a recursive logic program to return the last element in a list from a single example that is not an example of the base-case of the recursion; it's a lot harder than it may sound).
Now, I had been able to dismiss the earlier criticism as sour grapes, but this one I couldn't get over because my system really couldn't deal with it. So I tried to figure out why- where was the error I was making in my theories? Because my theoretical results said that my system should be able to learn that. Long story short, I did figure it out and I got that example to work, plus a bunch of other hard tests that people had thrown at me in the meanwhile. So I improved.
I still think my colleague's behaviour was immature and not becoming of a senior academic- attacking a PhD student because she did what you 've always done, beat your own system, is childish. In my current post-doc I just published a paper with one of our PhD students where we report his system trouncing mine (in speed; still some meat on those old bones otherwise). I think criticism is a good thing overall, if you can learn to use it to improve your work. It doesn't mean that you'll learn to like it, or that you'll be best friends with the person criticising you, it doesn't even mean that they're not out to get you; they probably are... but if the criticism is pointing out a real weakness you have, you can still use it to your advantage no matter what.
But above I'm only discussing my experience of criticism as an aside, unrelated to Gwern. To be clear, my original comment was not meant as constructive criticism. Like I think my colleague was at the time, I am out to get Gwern because I think, like I say, that he is a clueless AI influencer, a cheer-leader of deep learning who is piggy-backing on the excitement about AI that he had nothing to do with creating. I wouldn't find it so annoying if he, like many others who engage in the same parasitism, did not sound so cock-sure that he knows what he's talking about.
I do not under any circumstances claim that my original comment is meant to be nice.
Btw, I remember now that Gwern has in other times accused me , here on HN, of being confidently wrong about things I don't know as well as I think I do (deep learning stuff). I think it was in a comment about Mu Zero (the DeepMind system). I don't think Gwern likes me much, either. But, then, he's a famous influencer and I'm not and I bet he finds solace in that so my criticism is not going to hurt him in the end.
So thanks for that, even though it's entirely unrelated to AI.
[0]: though I've seen videos of the exact same effect on a plastic water bottle, but my brain didn't make the connection
An example of evidence against the reliability of educational attainment and intelligence heritability statistics: comparing intra-family heritability (across large numbers of families) to population-wide studies: for educational attainment, it turns out there's little correlation between the two; for simpler phenotypical traits, there's almost 100% correlation.
To sum this up:
1. The 2018 Plomin study gives sharply lower genetic/EA numbers than were floating around previously (say, from the Jensen-ist era)
2. Plomin's own numbers were preliminary and overstated
3. Researchers in the field criticized that study nonetheless
4. Subsequent studies on direct heritability and molecular heritability put even lower ceilings on it (basically, all credible behavioral trait heritability work has been done after 2018 --- and in fact this is broadly true of a lot of genetics work, not just trying to statistically mine behavioral traits out of genome scoring)
5. Even those results have flunked basic sanity tests (for instance, getting wildly different results in intra-family vs population-wide studies).
It's not looking good for people fixated on this idea.
† I'm being very loosy-goosey with the numbers and units here
We might want to look at the fundamentals: How is IQ qualitatively different from height, eye color, schizophrenia -- or any other highly complex, heritable polygenic trait? (One could also extend this to the many traits that animal breeders keep an eye on.) None of them have been fully pinned-down yet, but I don't believe that they can't be fully described in principle.
It's true that GWAS for intelligence explains <5% of variance today, but GWAS for height was in the same position a decade ago. Today polygenic scores for height predict over 40% of variance.
All this happens before we even reach questions about test-test reliability of IQ, or of whether gene-environmental interactions are uniform between Europe and other population cohorts (they do not appear to be!). It defines away SES confounding (which appears to be a significant issue). It has thus far largely ignored epigenetics. And, of course, for it to mean anything, the hypothesis also has to defend the idea that IQ/EA, at least in its genetic component, is immutable.
All that aside, I'm mostly just here to say that simple heritability statistics don't say what people on HN seem to think they say.