AI Tools Are Making the World Look Weird
Posted4 months agoActive3 months ago
strat7.comTechstoryHigh profile
calmmixed
Debate
70/100
AI BiasCultural HomogenizationLLM Limitations
Key topics
AI Bias
Cultural Homogenization
LLM Limitations
The article discusses how AI tools, particularly LLMs, may be biased towards Western cultures, and the discussion explores the implications and potential reasons behind this phenomenon.
Snapshot generated from the HN discussion
Discussion Activity
Very active discussionFirst comment
1h
Peak period
55
6-12h
Avg / period
17.8
Comment distribution160 data points
Loading chart...
Based on 160 loaded comments
Key moments
- 01Story posted
Sep 18, 2025 at 6:27 PM EDT
4 months ago
Step 01 - 02First comment
Sep 18, 2025 at 7:53 PM EDT
1h after posting
Step 02 - 03Peak activity
55 comments in 6-12h
Hottest window of the conversation
Step 03 - 04Latest activity
Sep 22, 2025 at 7:59 AM EDT
3 months ago
Step 04
Generating AI Summary...
Analyzing up to 500 comments to identify key contributors and discussion patterns
ID: 45295794Type: storyLast synced: 11/20/2025, 8:56:45 PM
Want the full context?
Jump to the original sources
Read the primary article or dive into the live Hacker News thread when you're ready.
The etymology was begging to be looked at, so I did. My understanding from undergrad was that it came from Greek "akrasia" as I recall but it's worth looking at:
https://www.etymonline.com/word/crazy
The use of language by leaders in the US lately sometimes seems reckless and inflammatory, even shocking and provocative, and I'll stop there before getting flagged for who knows what.
Poorly considered automation can create frictionless experiences for some and Kafkaesque experiences for the rest, where systems refuse to accept your atypical name, your atypical style of speaking is flagged as an indicator of fraud, etc. Automating processes involving people necessarily makes assumptions about those people, and such assumptions are often brittle.
For example, it's easy to imagine a resume filtering AI being implicitly prejudiced against people from Fictionalstan, because it was only trained on a few resumes from Fictionalstan and most of those happened to be classified as "unqualified". This is a danger anytime you have a small number of samples from any particular group, because it's easy for small sample sizes to be overwhelmed by bad luck.
In general I think these types of issues are best viewed as software bugs. It's a clearer and more actionable perspective than as ideological issues. If the software isn't serving some of our end users properly, let's just fix it and move on.
Read it or don't, it's your business, but if I ask you for evidence from the article and your response is to try to put words on my mouth (on an irrelevant tangent, at that), I'm going to write off your argument entirely.
However. No one gave them the right to speak for everyone, and in regular parlance “weird” isn’t a super nice thing to say about someone. They could’ve chosen something else like WESTED “Western, Educated, Stable, Technologically-advanced, Economically Developed” but didn’t. And they don’t get to choose how people will react when they’re called weird.
[1] https://news.ycombinator.com/item?id=45298741
There is too much victim mentality going on nowadays, and we continue rewarding that behavior so I am not surprised.
I'm a Caucasian living in Asia and the facial recognition systems that they recently required all banking apps to use struggles massively with my face.
I fully agree with you here that this isn't systemic racism, it's just a bug. It only becomes racism if they don't put any effort into fixing it.
By calling it WEIRD, the author is trying to drive home the point that the vast majority of people in the world are NOT in that culture that many westerners feel is 'normal', which would make it 'weird' in the sense that it isn't actually the norm.
Now, I have a lot of problems with the book and his arguments, but I don't think there is anything sneaky or nefarious about the word choice, it is very up front and straightforward as to the reasoning behind it.
The authors had a perfect opportunity to use Chinese models to see if their trend held up. Instead, they treated ChatGPT as the “default”. Sound familiar?
Possible that they're using different sources of feedback for different training though
There are parts of the world where constant person-electronic connection isn't a thing. Is that your point?
My anecdote is that before LLMs I would default to search Google in English instead of my own native language simply because there was so much more content in English to be found that would help me.
And here I am producing novel sentences in English to respond to your message, further continuing the cycle where English is the main language to search and do things.
But my guess is that the data sets used from the other languages are smaller (and actually, even if it had perfect access to every single piece of data on the internet, that would still be true, due to the astonishing quantity of English-language data out there compared to the rest. Your comment validates that). With less data, one would expect a poorer performance in all metrics for any non-Anglophone place, including the "cultural world view" metric.
It is such an eye-opening piece that explains so much of the world around us. He’s an anthropologist that goes into the psychology of it all. Touching on points like how religion plays a part in shaping the America of today and even how humans are worst at discerning faces today because we need to discern letters and words and dedicate brain power for that.
There are so many interest studies mentioned there, one that really stuck with me is how Protestant-raised Americans will work harder for the next day after having (reasearch-led) incestuous thoughts when compared to Catholics and Atheists.
He explains how monogamy is to blame for a lot of our western views today, and how Mormon towns in Utah were affected by not having monogamy as the basis of society (women there tend to prefer to be 2nd wives of a better man rather being the only wife of a lower-ranking man).
One of the wildest claims in there is the one that the north of Italy is more developed today because it was part of the Holy Roman Empire while the south wasn’t. About a thousand years separate these and he finds effects still. Mostly in connection to the spread of read/write to the public being a core tenant of Protestantism.
Anyway, this is not a summary of the book but instead a few points from it that really stuck with me after reading it. Fascinating stuff
I would note that the north and south of Italy have very different geography and climate. Which can be upstream of all sorts of things, culturally. The geography of Italy's two halves support different types of economic activity; and the social realities of living within these different economies, naturally evolves into major differences in culture. (Compare/contrast: the differing cultures of coastal vs midwestern America. Now imagine that split with a few thousand more years for the divergence to take hold.)
History happens once; but geography is always affecting a nation, all throughout its evolution. So if you're looking for reasons that two sub-populations within a country might have noticeable differences today, differing geography is going to be the "horse", while history is more of the "zebra."
That being said: geography can also constrain history.
Southern Italy is almost entirely coastline, in a part of the world where, for much of the last ~2000 years, everyone was constantly invading everyone else by sea. Northern Italy was relatively-more immune to amphibious assault, as its capitals could be situated more inland. (Rome itself — the exception that proves the rule — was located in south Italy, but was defended from amphibious assault mostly by the Roman Empire's huge naval home-fleet being docked to the southern-Italian coast; not by anything inherent to its location. Once the Roman Empire itself went away, big rich cities in southern Italy suddenly became juicy targets for conquest and/or sacking.)
This is wrong empirically and providing proof for this is how Acemoglu and Johnson won the economics Nobel. In basically all maps of voting patterns within Europe you can read its institutional history. You can see the border of the Holy Roman Empire in economic and voter data in Poland, you can see the iron curtain in every map of Germany.
If you want one of the counterexamples to your Italy theory, Venice was one of the richest middle-age cities in Italy and it is famously built on water.
You really have to explain specifically what you mean by this phrase, or else it's typically just saying you don't actually understand the rule or the exception.
It sounds like you're claiming Rome succeeded for reasons that overcame its geographical disadvantages, and due to this growth protected itself from naval invasions. But Rome was not a maritime power during its early republic period, let alone earlier. So why didn't Carthage or anyone else just sail upriver (Rome was not on the coast, just to clarify the context) and destroy Rome? How did Rome succeed in the first place to become a maritime power capable of defying southern Italy's geography?
[0] https://en.wikipedia.org/wiki/Unification_of_Italy
> monogamy is to blame for a lot of our western views today
Does the author use the word "blame" to mean "the reason for" or do they present it as a critique of monogamy? Not a big deal, just made me curious when I saw that.
https://pioneerworks.org/programs/julian-brave-noisecat-we-s...
I have absolutely nothing to back this up, but my gut tells me this risks being one of those bold claims that grows legs and runs for a while until we debunk it.
I enjoyed some books that don’t have anything unknown in its parts but that brought a lot of shift in perspectives for me, such as “Man’s Worldly Goods”[1] and “The Drunkard’s Walk”[2].
[1] https://www.amazon.com/Mans-Worldly-Goods-Wealth-Nations/dp/... [2] https://www.amazon.com/Drunkards-Walk-Randomness-Rules-Lives...
https://askakorean.blogspot.com/2013/07/culturalism-gladwell...
For example, suppose you sampled a group today and found an inverse-correlation between "good at recognizing many faces" and "good at recognizing written text"... That still wouldn't show that one facility grew causing the other to shrink, because maybe people are just born (or early-development-ed) with a certain bias.
https://www.unicog.org/publications/1-s2.0-S1364661311000738...
It would be ethically difficult to randomly assigned children to groups (A) taught to read versus (B) forced to remain illiterate while ensuring both groups had the same number of people's faces in their social circles.
Maybe, but e.g. Millikan's prize for physics was on the basis of results that appear to have been at least partially fabricated.
You can find, maybe, three or four such recipients out of 100. And they usually did make peace, even if they previously or later made mistakes.
On the other hand, does economics have less of a replication issue because it’s basically unreplicable?
On the more productive side, this suggests we might develop standardized tests of human capabilities and limits that would allow people of the future to compare themselves to us.
[1] https://www.unicog.org/publications/1-s2.0-S1364661311000738...
https://en.wikipedia.org/wiki/Visual_word_form_area
https://en.wikipedia.org/wiki/Fusiform_face_area
Dyslexia seems to be tied to some broader visual processing issues, which impair the ability to discriminate faces somewhat. But not the other way around.
If the two skills were strongly related, you'd expect a very strong and obvious link. Maybe in form of both performing poorly, if damage to the same pathways impairs both. Or as one performing poorly while another performs unusually well (super-recognizers? children who learn reading at 2?) - if the two skills compete for brain real estate and create a performance tradeoff, as claimed.
This seems like a bit of a tautology; how are they 'ranking' these men?
And probably could change based on the roles of first and second wives and, yes, how male status plays out and how it influences the life of the wife.
We don’t have legal polygamy but in many places there’s not much stopping people from living in an unmarried multi-woman household with a man (or vice versa). But it’s not a very common arrangement, and it’s interesting to think about why.
But it's fascinating to think about the second level "why": what made people encode monogamy and heterosexuality into their cultural canons (including their mainstream religion)? Was it property and rules about property? Was it to maximize the number of children, so that the group/tribe/kingdom would be militarily stronger than the neighbor? Or maybe it was to prevent some sort of very specific and concrete problem, real or perceived, that arose from tolerating free love, and that we today have no clue about?
I dunno about heterosexuality being encoded[1] into cultural canons, but for monogamy it's actually quite simple: violence.
Do you really want half your testosterone-fueled 18-28 year old males unable to attract a mate? There'll be continuous fighting to kill of the excess males.
===========
[1] As far as heterosexuality goes, it's not "encoded by wilful intention" so much as "this is the default". IOW, most people are happy going with the default, so if you make something opt-out, the majority won't opt-out. Same for opt-in. This is why countries that have opt-out organ donors have more organs donated, while countries that have opt-in organ donations have a fraction of he opt-out countries.
Defaults matter.
This is nonsense. Non-monogamy is relinquishing exclusivity. If a man can have multiple women, but a woman can't have multiple men, it's just a different form of oppression.
Monogamy is possessiveness, and possessiveness is what drives violence.
There are reasons to allow only one of the sexes to have multiple sexual partners/spouses.
* In a community with such liberal sexual practices, STDs spread more easily, especially in earlier centuries.
* It makes marriage intrinsically more complicated simply because of the more complex interactions. For example, if Alice is married to Bob, who is married to both Alice and Carol, who is married to both Bob and David, what are Alice and David to each other? Anything? Nothing? Is the entire married community a distinct entity?
* Relatedly, how is inheritance handled if such complex spousal organizations are going to be legally allowed?
You can use a condom. TIL, rubber condoms are a mid-19th century invention; a significant upgrade over sheep gut.
The alternative is called polyfidelity.
> [...] what are Alice and David to each other?
They're called Metamours.
> Anything? Nothing? Is the entire married community a distinct entity?
It's called a polycule.
> Relatedly, how is inheritance handled if such complex spousal organizations are going to be legally allowed?
You write a will.
By the way, inheritance laws are messy already as they are. Try figuring out how to reject inheritance (e.g. of debt) in your jurisdiction.
> There are reasons to allow only one of the sexes to have multiple sexual partners/spouses.
Yes, the reason is to reinforce division and oppression. One "side" is underprivileged, the other has to fight each other for supremacy. The stronger few win, everyone else loses. History is littered with examples.
Don't get me wrong, these are all very good questions. But we've figured all of these things out quite a while ago. People do live like that, and form lasting, loving communities. I'd wager that an entire society built on top of that would have no lesser chance at thriving than the one we've been born into.
Sure. Now. But monogamy and polygyny are a little older than condoms.
>Metamours [...] polycule
You're answering rhetorical questions which, incidentally, are not about terminology, but about legal and social mechanics. Knowing what a "metamour" is, says nothing about what the formal and informal responsibilities of the parties involved are or should be with respect to each other. My whole point is that not having to define such relationships and their expectations is a reason to forbid them culturally.
>You write a will.
How did that work before most people knew how to write?
>By the way, inheritance laws are messy already as they are.
That's not an argument in favor of legally legitimizing polycules.
>the reason is to reinforce division and oppression
I mean, I gave several reasons why historically either monogamy or asymmetric polygamy would have been preferred over symmetric polygamy, that have nothing to do with oppression.
>I'd wager that an entire society built on top of that would have no lesser chance at thriving than the one we've been born into.
Sure, maybe. Personally, I'm more of the opinion that cultural features are memetic, and that memes are not uniformly successfully propagated. If monogamous and polygynous societies are more common than polyandrous and polycular societies, it's probably for a reason.
Yes, that's what I've tried to imply. You name things, so you can discuss them in more abstract terms, so you can form a social & legal framework around those concepts.
> My whole point is that not having to define such relationships and their expectations is a reason to forbid them culturally.
>> [...] not having to define [...] is a reason to forbid [...].
Suppress the concept. "We don't talk about that."
> How did that work before most people knew how to write?
How did people enter agreements?
> Personally, I'm more of the opinion that cultural features are memetic, and that memes are not uniformly successfully propagated.
Agree. It's also how dictatorships rise. Another form of oppression that concentrates power and fires back at the group who have initially supported it. Another lose-lose.
Societies often overoptimize for a local maximum.
Only if your argument is that this behaviour is nurture, not nature. IOW, if your argument is "this behaviour is completely disconnected from instinct and has nothing to do with evolutionary pressure", then sure, your argument makes sense.
Many of the great apes, and indeed, other animals, don't think in terms of political soundbites, though, so we can readily observe that the behaviour "violence over mating rights" is a thing that developed in those creatures that eventually evolved into other creatures which evolved into pre-hominids which evolved into hominids which evolved into us.
Some things are instinct. It's a very large stretch to claim that violence over mating isn't instinct, but political.
Extraordinary claims require extraordinary evidence, and all that...
Violence about having an exclusive mate could be a purely cultural construct, reinforced from childhood. I can say that I always, personally, considered the idea of controlling a person abhorrent, as much as I found uninteresting the idea of orbiting my intimate life around a single person with special and very high privileges over whatever I do and think, including outside of bed. However, all my relatives were very insistent that I should date girls and marry. 100% culture, 0% nature. Of course, this is just a single data point.
The heterosexuality part being the "default" is a bit naive, because it ignores the lengths to which some people go to force their kids to be heterosexual. Again, anecdotally, my father sired two gay sons, who had to go to great lengths to have a less traumatic life. Sometimes I suspect my father wasn't that hetero himself, and was only ensuring the next generation inherited his cultural legacy/trauma.
I could consider an argument that a majority (heterosexual) imposes a cultural canon on a non heterosexual minority. But the problem with that is that we don't really know if that majority/minority split would exist without the very strong cultural conditioning. And, as I said before, I don't really believe that homophobia is something the Canaanites invented out of spite. Most likely, it was a cultural trait that conferred advantages to groups, particularly after the agricultural revolution locked human population in a cycle of growth and war for land--but that's just a pet theory of mine.
There are even theories (read the controversial book "Sex at Dawn" if you want the details) that our current cultural canons about sexuality run against what was our nature for hundreds of thousands of years.
"Could be" is not "probably is".
All the observable evidence points to "fighting for mates being purely instinctual".
The opposite, is in fact, true: not engaging in violence for mates is, in fact, purely a cultural construct.
Culture enforces that, via laws, norms, etc.
up until just 2-3 centuries ago, people did not think about these questions in a self-deterministic way, i mean they trusted the ancestors to sorted out most of the things over the centuries, and they were not suposed to take huge social/cultural reforms, only minor adjustments. did not even have so huge view on the spacial and temporal panorama of different cultures and societies over the whle word like us today. today even just binging up this statement makes people angry like "oh those silly old people. they _unconditionally_ obeyed to whatever their parents and superiors told them. this is the way to opression and tiranny, etc, etc. not like us! we are truely grown ups today. humanity is out of the dark child days. now! now disagree to _everything_. default is «i deny»". i mean … who do you think want to fool you? i think, on average, parents wants to leave the well-tested and proven-to-be-stable fundamental ideas about the world to their children. and those dont change very often; the more fundamental the less changing world-properties are.
but changes in society started to accelerate, so came social and economic revolutions which all want to redefine as much as possible. with really big improvements in the sociology, antropology and other culture-related disciplines, people started to believe that we are watching ourself from the outside, so able to manipulate the norms and the law to "make society better". only noone has the same definiton what is "better".
> This seems like a bit of a tautology; how are they 'ranking' these men?
I am not seeing the tautology. Can you explain?
> women there tend to prefer to be 2nd wives of a man with more than one wife than being the only wife of a man with a single wife
How is that statement tautological?
I don't think it is? A priori it's not at all obvious which option women would be expected to prefer.
I've noticed over the years many chains of reasoning - made up of what I believe someone called "cocktail party" pithy takes - that only last as long as you don't dig into the nuts and bolts of them. Pleasant little takes on our psyche and behavior that makes for nice reaffirming thoughts of our views but break down under later analysis.
It feels like we have sometimes accreted an amalgam of these pithy takes based on very small, one off, studies (never replicated) that let us comfortably assemble an affirmation of our broader takes.
This is a rotten thing to say about your book recommendation, given I have never read it (I hope you'll forgive me), but based on the last few years of the replication crisis, do you think, in your heart of hearts, that what you are describing truly does stand up?
Here are the cocktail ideas. Hits the spot.
it is a common rhetorical device to phrase something as an absolute when the negation of it is only an edge case.
hence
>Hillel interviewed a civil engineer who said that they had to move a bridge! Of course, civil engineers don't move bridges as frequently as programmers deal with changes in software but,
The analogies seem to just be missing the point. There's constraints, so what?
I've worked in hard science, engineering, and software. No one is omniscient, so the goals evolve and pivot during the project. That's pretty standard practice. You can't just plan and execute unless you're omniscient. Honestly, the big differences I see is that programmers spend less time at the drawing board and engineers and scientists spend much more time there because working in physical space is very costly and time consuming. But there's a lot of similarities. Programmers would be more effective if they spend more time at the drawing board and engineers would be more effective if they could hack on their tasks more cheaply (which is why sim has had such an impact for them)
Would they, though? As you've correctly pointed out, design goals in software engineering get shifted by decision-makers because its cheaper than in civil engineering. The whole point of the ToP article is pointing out that software engineers have to account for possible future radical changes that in other branches of engineering are at most exceedingly rare. Any time you spend on initial planning beyond a bird's eye view may be time wasted.
But think about it this way, how do you plan a vacation? I'll tell you how I do it and you'll tell me if you're different, which is okay. There's no "right" or "wrong" way. I'm sure some things will be different and it's going to change every vacation, but bear with me here, since this is more of a communication aid than telling you how to vacation lol.
Prior to the vacation I plan out the major things, like how do I get there, how do I get back, the lodging, and so on. I'll have some key things planned out that I want to do. But I won't ever have everything planned out in detail. I actually do not like having each day scheduled unless that is more tentative and and acting as a stand in. Then after traveling my schedule changes, especially in the beginning. Things are different than I expected, so I'll learn that I'd have more fun doing X instead of Y. Or I find that I really like Z so I want to allocate more time to that. Maybe the weather changed and so I can't do P, and I instead do Q. I'll ask locals and hotel staff what their favorite places are to eat and go there. I'll likely have had a few more famous places to eat laid out, but definitely not every mean. Fuck, some days I'm just tired and would rather call the day early and do takeout. As the vacation closes, things become way more "stable". If I go to the same place in a second vacation I'll definitely lean on my experiences and do things very differently, usually with less flexibility (depending how much I was ale to discover what I like doing the first time around).
The point is that no matter what you're doing, there is exploration and exploration is coupled with the doing phase. It'd be pretty fucking exhausting to plan out the vacation at the airport. I mean people do do this and I'm sure you could still end up having a great time. But a little planning can really go a long way, right? That's the planning. Just like when you get back to your hotel at night and modify plans. That's a planning stage too. The logistics of a vacation almost force this kind of behavior on people. But in programming it is much easier to pivot, almost to the degree that you can be mid meal at a restaurant and decide you want to eat somewhere else. Being able to pivot like that is an incredibly powerful and useful feature, but this doesn't mean that planning still doesn't provide major benefits. Going in blind is crazy! If anything, it makes it more important. In both physical and software you still are time limited and unable to brute force all paths. In physical you can't jump mid meal and even if you pivot as soon as you get a good look, you're much more limited to what you can pivot to because you can't teleport across town. But in programming, you can. You can brute force sometimes, but that clock still ticks forward and you're still going to benefit from planning. The real difference is in physical I might be able to consider 2 dozen places to eat but in software I might be able to try a few hundred. Still need to plan if there's a few thousand, right?
You need to balance these things: the planning, exploration, and execution. Working in physical forces a dominating planning stage and more careful exploration stage, because execution is so costly. But in software execution is cheap. That doesn't mean we should throw out the planning stage, it means we can exploit it much more effectively!
In programming, engineering, and hard science your goals evolve during development. There's always discovery during the doing process that necessitates pivots, and sometimes hard pivots. The main difference I've seen is just how much time goes into planning. Software has an advantage in that when working with physical things mistakes are incredibly costly both in money and time. You fuck up a tolerance and you might need you wait a few weeks for the part to be remade and you might have an expensive paperweight (hopefully you can use for some testing).
So what that leads to is more planning stages. That's not just make a plan and go, but make a plan, go, regroup, replan, go, repeat. It often means gathering people who are the owners of different parts of a project because you can't just duct tape things together and the most permanent solution is a temporary solution that works. This greatly affects how I go about programming and is something I notice I do differently from my peers. I spend a lot more time at the whiteboard while most people I know never visit one. I'm not spending all my time there, or even most, but I couldn't do my job without "pen and paper".
In programming the "laws of physics" aren't constantly changing and you're not "building a plane while flying it" (how would it even get off the ground lol), but your requirements are constantly changing. That's... normal engineering and normal in both experimental and even theoretical science. That's because we're not omniscient and you don't know the full answer from the get go lol.
This isn't to say in trad engineering and science we doing also "move fast and break things". Just like in programming you'll build toy models or scaled down versions. But I do think programmers could benefit a lot more and make a lot fewer mistakes (substantially reducing future workloads) would they spend a bit more time at the drawing board. It's great that in programming we can jump in and poke around and experiment so much faster than the physical world allows us, but it seems that this feature is overused instead of being used in addition to planning and designing. That's what actually made me come over to this side, was the ability to iterate faster. But sometimes you gotta take a step back and look at things. Sometimes you gotta move the bridge. Sometimes you gotta tear it down to build an entirely new one. The latter is actually much easier in programming and honestly I feel like it's used less frequently. But that's like being unwilling to throw away your first draft when writing a report (or anything). Why hold on? The first draft's job is to get the stuff out of your head and see it in a more physical form. It's so easy to rebuild, magnitudes easier than doing it the first time, but it always hangs on as if losing it is losing work.
I can tell you from my personal experience that the info there has helped me understand the differences between how people think in Brazil (where I come from) and how people think here in the US. Could it be me pattern matching? Possibly
I wouldn’t expect all of it to be true, but I would be very surprised if most of the sources the author provide are false or lack theory and tests, since he explain control groups and experiments in details.
I’m not that married to the book either, as I find some claims rather bold (like the Italy divide)
The title does sound catchy tho
Edit; the author’s main point is how the papal rule on monogamy changed Europe and its colonies to this day, which I didn’t capture on my main comment. Lots to unpack there
That's not the issue. The replication crisis is the phenomenon that many scientific results and conclusions which originate from serious, peer-reviewed research, couldn't be replicated by other researchers, and sometimes not even by the original scientist. This is especially concerning because many results with strong statements – unintuitive ones as well as bias-confirming ones – turned out to be non-existent [1]. It doesn't necessarily have anything to do with "shallowness" or "cocktail-party takes", although the strength of the purported effects, combined with pop-science simplifications and reductions, lend themselves well to such memetically spreading factoids.
[1] The "softer" sciences tend to be much more affected by this than the "harder" sciences.
This seems to happen generally for two reasons: Even a neutral comment is evaluated for what stance it most closely aligns with, and then downvoted to suppress the opposing view just in case. Or alternatively, a comment that appears low-key combative (but really isn't directly so) gets downvoted in an attempt to ensure harmony.
Both moves to me have "culture war vibes", and come from either adopting those habits or feeling very tired from strife.
I think it's increasingly easy to fall into either bracket, but let's not do that on HN! If a comment is generally polite enough, the only bar to meet is adding new information or new thought into a conversation. None of us come here to be pandered to, and getting challenged by viewpoints that force you to consider the corner cases of your own views is half the fun.
The question asker doesn't know the work, doesn't respond to what OP said, but instead challenges OP about methodology, and leads with 'I know this question can seem combative'. Then falls to 'do you think, in your heart of hearts'... how does that question and asking for assessment align with the 'I'm all about methodology' stance of the question asker? It sounds a like a 'core values' assessment/assignment not a 'the room for error in this study' assessment is being asked for. The question on the whole:
'I know this can sound combative but I'm just asking questions. Given other things have been bad and knowing nothing of this being talked about, but pointing out it's probably completely wrong (based on nothing but X other thing is wrong).... really, in your heart, do you believe you are coming from integrity?'
That sounds toxic AF.
Personally as a relative newcomer here it seems like there is a lot of this 'just asking questions' on HN.
Edit: Throttled. I pointed out how I saw the post could be (mis)interpreted. Yeah, that necessitates me replaying it back how it could be (mis)interpreted. That is valid when my point is about... how posts could be (mis)interpreted resulting in a poorer quality of discussion. Sorry if you didn't understand the point I was trying to get across. I didn't say the interpretation was valid, I said here is how posts like the one the person I responded to referenced can derail discussion in an era of 'just asking questions'. Zero disingenuousness nor unconstructive on my part and it's wild you can't see that. My post was about better quality discussion using the message the person I responded to used. Yours is about calling me specifically out. Which is more 'constructive'?
That's not the case here. Non-replicable results from studies in the social sciences are a very real, very frequent phenomenon, and the first question to ask when seeing a claim about a significant effect should be "Has this been replicated?". Being sceptical (without being overly negative or critical) is not "toxic" as you call it, instead it protects us all from becoming trapped in our bubble.
> 'I know this can sound combative but I'm just asking questions. Given other things have been bad and knowing nothing of this being talked about, but pointing out it's probably completely wrong (based on nothing but X other thing is wrong).... really, in your heart, do you believe you are coming from integrity?'
> That sounds toxic AF.
You had to rephrase PeterHolzwarth's post and put word in their mouths to make it sound "toxic AF". That's a disingenuous and unconstructive thing to do in a discussion.
Example: Two people with similar classically liberal values hear the same "pithy take" on a politically contentious issue. One accepts it as presented, the other digs in and finds it doesn't hold up to scrutiny at all.
Almost invariably, the skeptic is ostracized, his findings met with incurious dismissal.
The rate of adoption has accelerated along with the news cycle.
The patterns are there and are hard to deny. The reasoning and explanations of these types of books? Don't take them for granted, do your own research if anything is of particular interest, think for yourself, etc. The books can be of value without being 100% correct.
But lots of otherwise good books have these little mistakes on them, so I find it best to gloss over them and see if the point stands without them.
The Italian states started becoming more autonomous in the mid-1100s. It wasn't until the 1600s that they were fully independent.
Must suck to be a lower-ranking man lol
I bet you think "2nd" means "secondary."
I bet the parties to the marriage think "2nd" means "most recently allocated."
On the one hand, this reflects the US dominant position in world affairs and the fact that probably most of the training materials come from there.
But on the other hand, there are some outlier results that are left unexplained. For example, ChatGPT is even more aligned with Japan than with the US.
I have no expertise in this field.
Is it actually even more aligned? Or is it simply aligned with the elements of Japanese culture and/or media that are exported to the West?
This has been true of web search since forever mind you. The wev has always been culturally delineated by language, and the English Web as I call it is not the only web.
I'm wondering the same thing, in addition to the related question of “Would an LLM perform better or worse if prompted with languages other than English?”.
(I did then go and check the book myself; ChatGPT in English was right, the name is there)
For example (not actual output):
Input: "こにちは"(konichwa) Qwen Thinking: "Ah, the user has said "こにちは", I should respond in a kind and friendly manner.
Qwen Output: こにちは！
It quiiiickly gets confused in this, much quicker than in English.
It was pretty good at one shot translations with thinking turned off however, I imagine thinking distracts it from going down the Japanese only vector paths.
That said, in my tiny experience, LLMs all think in their dataset majority language. They don't adhere to prompt languages, one way or another. Chinese models usually think in either English or Chinese, rarely in cursed mix thereof, and never in Japanese or any of their non-native languages.
So all information gleaned reading a glyph in the context of japanese articles would be totally different vectors to the information gleaned from the same glyph in Chinese?
Why?
Because the creators want the reasoning trace to be human readable. And without a pressure forcing them to think in English, they tend to get weird with the reasoning trace. Wild language-mixing, devolved grammar, strange language-mixed nonsense words that the LLM itself seemingly understands just fine.
In other words, is the issue in the defaults or is it impossible for AI to respond from other cultures?
31 more comments available on Hacker News