AI tools are making the world look weird • Hacker News Board

Re: AI tools are making the world look weird

#101

Post by ACCount37 » Fri, Sep 19, 2025, 9:00 AM UTC

Waterluvian wrote: post #24

psidium wrote: post #5

Can’t recommend the book that coined this acronym enough: The WEIRDest People in the World Book by Joseph Henrich. 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 discer…

> humans are worst at discerning faces today because we need to discern letters and words and dedicate brain power for that. 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.

There are people who have face-blindness (inability to discern faces) or dyslexia (visual processing disorder that leads to severe difficulties with reading). The two aren't strongly correlated.

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.

Re: AI tools are making the world look weird

#102

Post by watwut » Fri, Sep 19, 2025, 9:02 AM UTC

YurgenJurgensen wrote: post #74

maxbond wrote: post #34

Earlier quoted context omitted.

WEIRD is not pejorative in TFA. There's no problem being WEIRD. I am WEIRD. What's alleged in TFA is that AI, as it's currently deployed, is implicitly chauvinistic towards perspectives other than WEIRD. This sort of thing has historically been a problem with AI/ML and automation in general. The classic example is cameras deploying autofocus features that fail on non-white faces (which has happened several times). Po…

When a member of a group says that a term used to refer to them is pejorative, is it now an acceptable response to simply say ‘no it isn’t’? That hasn’t been the case for decades.

You know, are you seeing it as pejorative or just want to silence the whole idea? Because people who came up with WEIRD acronym were ... WEIRD. They were literally talking about themselves and about weakness of studies that were made.

Re: AI tools are making the world look weird

#103

Post by numpad0 » Fri, Sep 19, 2025, 9:04 AM UTC

ehnto wrote: post #49

tropdrop wrote: post #47

Earlier quoted context omitted.

ChatGPT is worse in Russian. Example: after accurately noting that a name appeared in a particular Russian book, it asked if I wanted the direct quote in Russian. I said yes. At this point it switched to Russian output but could no longer find the name in that book, and then apologized for having used what seemed to have been "approximations" about the book before. (I did then go and check the book myself; ChatGPT in…

I was using Qwen3 locally in thinking mode, and noted that even if it is talking to me in Japanese, it is doing it's "thinking" steps in English. Not having a full understanding of how the layers in an LLM handle language connections I can't say for sure, but for a human this would result in subpar outcomes. For example (not actual output): Input: "こにちは"(konichwa) Qwen Thinking: "Ah, the user has said "こにちは", I shoul…

I'm kind of wondering when will it become a universal understanding that LLMs can't be trained with equal amounts of Japanese and Chinese contents in training data due to Han Unification, making these two languages incoherent mix of two conflicting syntax in one. It's remarkable that Latin languages is not apparently facing issues without clear technical explanation as to why, which I'm guessing has to do with the fact of granularity of characters.

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.

Re: AI tools are making the world look weird

#104

Post by Terr_ » Fri, Sep 19, 2025, 9:09 AM UTC

suddenlybananas wrote: post #98

Terr_ wrote: post #35

Earlier quoted context omitted.

I'm struggling to think of any way to test the hypothesis which is (A) practical and (B) accurate. 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.

The part of the brain that recognises faces quite literally shrinks in literates compared to illiterates https://www.unicog.org/publications/1-s2.0-S1364661311000738...

I'm quite willing to believe that human brains do Weird Stuff with respect to reusing circuits and development, but I still don't see how they concluded "the skill encourages the brain state" by disproving "the brain state encourages the acquisition of the skill."

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.

Re: AI tools are making the world look weird

#105

Post by psidium » Fri, Sep 19, 2025, 9:15 AM UTC

PeterHolzwarth wrote: post #53

psidium wrote: post #5

Can’t recommend the book that coined this acronym enough: The WEIRDest People in the World Book by Joseph Henrich. 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 discer…

I have to ask (and I don't mean this combatively) - given the ongoing realization of the replication crisis, how likely is it that the book you mention reflects a summation of the "too pat" studies about human behavior that, en masse, always seem pithy in an interesting headline, but years later end up being completely bunk? I've noticed over the years many chains of reasoning - made up of what I believe someone call…

I do not have the time now to craft you a full answer as I don’t have the book on hand and have only been commenting from memory so far. But to give you a quick answer: I don’t think all of it is shallow, especially given the format: the book is mostly a prose re-writing of the author’s own peer-reviewed anthropology scientific papers. Most of the authors claims are backed by actual papers for reference on the footnotes. As for replication it seems that the author himself replicated some of his studies with different hunter gatherer societies in the world. It’s been a good while since I read it.

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

Re: AI tools are making the world look weird

#106

Post by psidium » Fri, Sep 19, 2025, 9:19 AM UTC

Terr_ wrote: post #35

Waterluvian wrote: post #24

Earlier quoted context omitted.

> humans are worst at discerning faces today because we need to discern letters and words and dedicate brain power for that. 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'm struggling to think of any way to test the hypothesis which is (A) practical and (B) accurate. 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.

The author uses mostly hunter-gatherers tribes/societies in different continents as the control groups, usually. Most of his work is in anthropology.

Re: AI tools are making the world look weird

#107

Post by ACCount37 » Fri, Sep 19, 2025, 9:22 AM UTC

ehnto wrote: post #49

tropdrop wrote: post #47

Earlier quoted context omitted.

ChatGPT is worse in Russian. Example: after accurately noting that a name appeared in a particular Russian book, it asked if I wanted the direct quote in Russian. I said yes. At this point it switched to Russian output but could no longer find the name in that book, and then apologized for having used what seemed to have been "approximations" about the book before. (I did then go and check the book myself; ChatGPT in…

I was using Qwen3 locally in thinking mode, and noted that even if it is talking to me in Japanese, it is doing it's "thinking" steps in English. Not having a full understanding of how the layers in an LLM handle language connections I can't say for sure, but for a human this would result in subpar outcomes. For example (not actual output): Input: "こにちは"(konichwa) Qwen Thinking: "Ah, the user has said "こにちは", I shoul…

Quite a few reasoning LLMs do reasoning in English only. Because the RL setup specifically forces them to do so.

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.

Re: AI tools are making the world look weird

#108

Post by ehnto » Fri, Sep 19, 2025, 9:26 AM UTC

numpad0 wrote: post #103

ehnto wrote: post #49

Earlier quoted context omitted.

I was using Qwen3 locally in thinking mode, and noted that even if it is talking to me in Japanese, it is doing it's "thinking" steps in English. Not having a full understanding of how the layers in an LLM handle language connections I can't say for sure, but for a human this would result in subpar outcomes. For example (not actual output): Input: "こにちは"(konichwa) Qwen Thinking: "Ah, the user has said "こにちは", I shoul…

I'm kind of wondering when will it become a universal understanding that LLMs can't be trained with equal amounts of Japanese and Chinese contents in training data due to Han Unification, making these two languages incoherent mix of two conflicting syntax in one. It's remarkable that Latin languages is not apparently facing issues without clear technical explanation as to why, which I'm guessing has to do with the fa…

Would they not quickly bocome divergent vectors? In the same way that apple and Apple can exist in the same vector set with totally different meanings?

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?

Re: AI tools are making the world look weird

#109

Post by psidium » Fri, Sep 19, 2025, 9:26 AM UTC

cjauvin wrote: post #23

psidium wrote: post #5

Can’t recommend the book that coined this acronym enough: The WEIRDest People in the World Book by Joseph Henrich. 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 discer…

One of my greatest pleasure of random walking the internet is building my list of possible next books to read.. thank you for this one!

Do you have some recommendations? I remember seeing Show HN tools gathering book recommendations from here, but I like specific experiences

Re: AI tools are making the world look weird

#110

Post by psidium » Fri, Sep 19, 2025, 9:32 AM UTC

lelanthran wrote: post #96

dsign wrote: post #51

Earlier quoted context omitted.

The first level "why" at least is straightforward: Christianism. Even when not directly imposed, it's still the basis of the Western system of values and morals. 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 chil…

> 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)? 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 continu…

Literally one of the main points of the book: the church’s mandate and enforcement of monogamy in Europe lowered violence so much that a different kind of society emerged.