stuckinbathroom
No bio...
User ID: 903
I admit I have never tried this, but I’ll take your word for it; it does seem plausible that Opus 4.6 or equivalent would be able to one-shot a simple program that computes the state of the board after a given sequence of past moves and validates that a proposed next move is legal.
Still, this raises 2 questions, one rather surface-level/product focused and one deeper and more architectural.
Firstly, why should I as the user have to prompt the AI to make a program to ensure that it doesn’t go off the rails? Why can’t it figure that out for itself? For example when I ask a modern SOTA AI to answer trivia questions, I don’t have to tell it to go to such and such website; I don’t even have to tell it to search the internet. It just “knows” without prompting that a Google search is the right tool for the job. Why can’t it do the same thing for chess? Or for that matter, for the old “number of Rs in ‘strawberry’” question that it kept stumbling on last year? There are any number of common natural-language queries that really boil down to a problem of logic or some other formal system—it should be the AI’s job, not mine, to identify them, come up with the right formalism, and then use it to solve the problem.
I suspect this shortcoming may be trivially resolved by adding something like “Always consider whether you can map this question, or some piece of it, to a problem that can be solved in Python and remember that you have access to a sandboxed Python environment” to CLAUDE.md or the system prompt or whatever. Fair enough. But this gets us to the second and more fundamental question: for a given AI and problem, it’s not always obvious what the best formalism or representation of that problem is. Let’s go back to the chess example; suppose the AI writes a Python program to keep track of the board state and ensure all of its moves are legal. On some level, the “game loop” then becomes something like:
- I type in a (legal) move in algebraic notation
- AI appends that move to a text file
- AI runs the program to print some representation of the board, after all the moves recorded in the text file, to its internal context
- AI decides on its move—either by simply treating the board state as another sequence of input tokens and emitting the corresponding output, or by running some other program of its own devising, but for the sake of discussion assume the former, as the latter presupposes the ability to one-shot Stockfish which AFAIK is beyond the current SOTA—and appends it to the text file
- AI runs the program again to confirm the move is legal; if not, erase the last move from the text file and goto 3
- AI prints its move from above to the screen so I can see it
- Goto 1
Let’s drill into step 3: what is the optimal representation of the board that the AI should be using for its own benefit? It’s a bit of a trick question: “optimal” here means something like “maximizing the probability of the AI winning the game” but perhaps also “minimizing the probability of making illegal moves which cause it to waste time looping through steps 3-5 again”. As a human I can certainly come up with various representations; an obvious one would be to render the board as an 8x8 CSV or Markdown table. But I have no idea whether this is “optimal” for the AI in this case, and in general I may not even know what “optimal” should mean. Again, it should be the AI’s job to figure all of this out—otherwise it’s not worthy of the name AGI in my book.
One last thing: I don’t actually care whether AI uses a sandboxed coding environment or whatever to solve problems. Perhaps it will turn out to be the case that just scaling up—more compute, more RLHF, bigger transformers, bigger context windows—will suffice to get LLMs to the point where they can (e.g.) play and win games they have never seen before purely by transforming tokens to tokens, without the use of external tools, or to the point of one-shotting algorithmic solvers like Stockfish. If so, great; one is reminded of the old Deng Xiaoping quote about the color of the cat. But based on what I’ve seen so far, it looks like there’s a ton of low-hanging fruit in the direction of “just use the tools already at your disposal” at our current levels of model complexity.
stuckinbathroom
2d ago
LLMs play chess entirely through text. It's the equivalent of asking a person to play a game of correspondance chess, buth they can't recreate the game physically, they can't have any drawings of the game, all they can do is have a record of moves already made. Outside of literal chess masters, how many humans would get through such a game without making a mistake?
But LLMs* also have a massive advantage over an unassisted human: they have access to the internet, or at least to a sandboxed Python interpreter or similar coding environment. So the fact that they are constrained to text-based I/O should really be no excuse: there’s absolutely nothing, in principle, preventing the LLM from “thinking” to itself “Hmm, I’m being asked to provide answers about a formal system. Let me create a computer program to record the state of the game and make sure I don’t make any illegal moves.” But current SOTA LLMs never think to do that, even when all the tools are at their disposal.
In other words, the equivalent human activity is not playing correspondence chess with nothing but a record of all past moves, but rather playing correspondence chess with a book of chess rules plus pen and paper (or text editor and Python interpreter, if you like)
*OK, I admit I am playing fast and loose with the definition of “LLM” here. In the very strict sense, language models do only transform one sequence of tokens into another, as you said. But in the colloquial sense, which is also the more relevant one for discussing the abilities of SOTA consumer-facing AI, “LLM” refers to a product like ChatGPT, Claude, etc. consisting of a core language model (in the strict sense) together with tools that it can invoke to solve problems.
stuckinbathroom
3d ago
IndianMaleChineseFemale couples
Is it possible to learn this power? Uh, asking for a friend, of course
stuckinbathroom
4d ago
Malaysia’s wealth and development is almost entirely due to the non-Muslim Chinese community.
The Gulf’s is due to oil money and incentives for Westerners to come and do business there.
Turkey’s is due to the uniquely strong tradition of secular, Westernized institutions, enforced by regularly-scheduled military coups whenever the backwater Islamist yokels start getting uppity.
stuckinbathroom
5d ago
Could probably do an interesting "What sort of Attack On Titan fan are you?" question
brb, asking ChatGPT to photoshop Jesse Plemons from the Civil War meme into a neckbeard weeb
stuckinbathroom
5d ago
lol no, the meme was meant to indicate that I spent too long in that hellscape (at least dating-wise) for crimes I didn’t commit
I did my time (twelve_years_of_it_in_azkaban.gif) in the Bay Area and things didn’t seem vastly different there. The biggest difference I noticed is that nerdy white male/East Asian female was always an extremely common pairing in the Bay but only recently (post-pandemic, roughly) became comparably prevalent on the East Coast.
I used to think it was purely a matter of the gender ratios: even a mid white guy in NYC can easily find (white) BPD art hoes, starving students, or fashion/publishing/journalism girlboss types, while mid white guys in the Bay are SOL without the boost they get in the eyes of East Asian women. But the gender ratios are the same as ever, if not even more lopsided than before, and yet WM(E)AF has only trended upward in NYC for the past half decade.
stuckinbathroom
5d ago
Precisely this. East Asian men, by contrast, have something like a -2 or -3 handicap on average* while the women have a +5 buff
*though these days it is very context-dependent; e.g., Korean men have a big advantage with white female K-pop fans
stuckinbathroom
5d ago
So yeah, I do think that the arranged marriage/matchmaking thing still exists for 2nd gen South Asian Americans to a far greater extent than for (e.g.) East Asians. This plays a big role in finding matches for the men, who are the least desirable demographic out there and face a terrible slog on the apps.
But it takes two to tango, as they say, and I suspect the primary reason why the women are also available to be matched in the quasi-arranged marriage market is that white men are much less eager to snap them up as compared to East Asian women. IME the rates of WMIF dating relationships and marriages are roughly the same, so it’s not so much that South Asian women are fooling around with white dudes in their early 20s before settling down with a parentally-approved Nice Jewish Indian Boy From A Good Family right as the Wall approaches. It’s just that white guys aren’t asking them out all that much in the first place, and from my (single straight male) perspective it’s pretty easy to see why: East Asian women have more of the traits that men generally find attractive, viz. petite and slim build even after marriage/children, fair-but-not-ghostly-pale skin, and lack of body hair/body odor.
I also don’t think that 2nd gen South Asian Americans are that much more susceptible to parental pressures re: choice of spouse than East Asians. In both cases, the parents would prefer that their children (especially daughters) marry within the race and grumble but usually grudgingly accept if and when that doesn’t happen; and in both cases, at roughly equal rates in my anecdotal experience, sometimes the children acquiesce and sometimes they go their own way. The degree of submission to the parents depends mostly on the strength of the relationship with the parents, which again I think is not vastly different on average between East and South Asians.
Though if I had to hazard a guess, I suppose I would agree that South Asians (even 2nd gen) tend to defer to their parents somewhat more so than do East Asians, and both groups defer far more so than do whites.
stuckinbathroom
5d ago
(I am a 1.5 gen immigrant to the US from South Asia)
Honestly I think this is not entirely due to South Asian insularity/clannishness. The bigger reason is that South Asian men are the least desirable on the dating market, plus in all races (except perhaps East/Southeast Asians), women generally prefer their own; moreover, white men seem markedly less likely to date and marry South Asian women as compared to East Asians. This nets out in South Asian men pretty much only being able to pull South Asian women, hence the phenomenon you notice.
I will admit, though, that South Asian parents/older relatives are much more into playing matchmaker for the younger generation than East Asian parents. Also, South Asian parents heavily stress the importance of marriage, especially for daughters, whereas East Asian parents stress the importance of being skinny/good looking and seem to be less vocal about marriage specifically (though of course they generally don’t want their daughters to end up as spinsters)
stuckinbathroom
7d ago
Further nitpick: despite that being how it’s written in Cyrillic, in modern Russian it is pronounced something like “kto kavo”
stuckinbathroom
7d ago
Can’t speak to her looks, but she would be at least a year older
This is like when Dems chased out Al Franken or Andrew Cuomo, there’s a noose tightening now around straight heterosexual male sex. I see a lot of conservacons cheering but this seems longhoused to me.
Al Franken perhaps, but I always got the impression that Cuomo’s cancellation was the left’s way of shitcanning him for bungling the pandemic with a minimum of cognitive dissonance. Much more palatable to run with the fig leaf of sexual impropriety than to recant the glowing hagiographies they had spent the previous couple years composing.
stuckinbathroom
8d ago
I (and they) would much rather go back to China
stuckinbathroom
9d ago
I believe the increased scrutiny ("Are you a chinese spy?") and increasingly hostile culture against China lead to many Chinese deciding to go back to their country.
Not to mention the near-impossibility of dating as an East Asian male fob in the US, especially in the major tech hubs (Bay Area, Seattle)
stuckinbathroom
10d ago
I remember one of the things that really struck me about Halo in its day was that it didn't have bosses.
Arguably hunters are mini-bosses, especially on the higher difficulties. And there was the unique elite with gold armor and the plasma sword—I think in “Silent Cartographer”—but he was easily dispatched with a single plasma grenade so it’s hard to call him a true boss.
Hungary is also an interesting sign of how this new "postliberal right" have abandoned the plot on traditional conservatism.
See, the thing is that actual traditional conservatism doesn’t necessarily have any strong ideological reason to join itself with pro-business, anti-regulation free market economics. The latter is approximately what Americans call libertarianism, and what Europeans call (classical) liberalism. It’s really a historical contingency that in America, this political strain happened to join forces with the religious right/moral majority/tradcon types from approximately the Reagan era to the Trump era; even then, it was far from a solid Republican voting base (cf. Clinton peeling some of them away with the Third Way, “the end of welfare as we know it”, NAFTA, etc.)
There’s no reason to think this should be a general law of conservative politics; indeed, globally speaking, it tends to be the exception rather than the rule, especially in systems that favor the formation of smaller, focused parties instead of two big amorphous tents. Hell, even within the FPTP Anglosphere, it’s not uncommon to find conservatives, leftists, and classical liberals form 3 entirely separate parties (cf., respectively, Tories, Labour, and LibDems in the UK; Tories, NDP, and Liberals in Canada)
Or when the pro-government media tried to "attack" him that in the EU parliament, while waiting for his turn to speak, he had his hand in his pocket and was adjusting his penis, or that the shape of his large penis is sometimes seen in his tight pants on some of his photos.
“My opponent has a magnum dong, for which he needs monster condoms” is … not the angle I would take for an attack ad. But what do I know; perhaps it comes across better in Hungarian.
stuckinbathroom
10d ago
Magyar knows the symbols and spefaks the language, and uses national symbols, national clothes, songs and so on.
Nominative determinism strikes again!
stuckinbathroom
11d ago
I'm admittedly fortunate in that nobody in either of my families is a high-grade fuckup and I could see how that'd cause issues with the current state of things.
IMHO this is literally the entire argument (whatever you think of its merits) for a redistributive welfare state to take care of the young, the old, and the incurably indigent. You are indeed very lucky to have at least some family members of an older generation alive and in good enough shape to provide a degree of childcare, but from behind the Rawlsian veil, such good fortune is hardly guaranteed.
stuckinbathroom
11d ago
I think the point that comparatively few people used to live long past the point of becoming a net fiscal drain still stands, though. Modern medicine has made eking out low-quality end-of-life years, at exorbitant expense to the taxpayer, much more common than it once was. See also the comment downthread about Down syndrome life expectancy over the years.
The funds were “invested” in Treasury bonds which means that they were spent more or less immediately on whatever the government at the time wanted. The return on that investment will ultimately come from taxing future generations, or inflating away the debt, so it’s really just an intergenerational Ponzi scheme with extra steps.
But I agree with you that this was the plan from Day 1, so strictly speaking it’s not a broken promise per se (at least not yet).
stuckinbathroom
11d ago
Maybe they should have a special express lane on the highway for people who pay more than $50k a year in taxes.
I’d settle for a plaque from Lockheed Martin/Raytheon/Northrup Grumman every time I pay enough tax to fund a new Tomahawk or something
stuckinbathroom
14d ago
… which incidentally is exactly why people were worried about JFK becoming the first Catholic president back in the day: would he refuse orders from Rome if his immortal soul were on the line?
relevant Far Side
More options
Context Copy link