Comment by vunderba | Hacker News Reader

Comment by vunderba | Hacker News Reader

Parent Full thread

I'm not sure how this affects the premise of the article, but the "jaw dropping quote created by an LLM about NYC" reads like so much pretentious claptrap:

  Because New York City is the only place where the myth of greatness still 
  feels within reach—where the chaos sharpens your ambition, and every street 
  corner confronts you with a mirror: who are you becoming?

  You love NYC because it gives shape to your hunger. It’s a place where 
  anonymity and intimacy coexist; where you can be completely alone and still 
  feel tethered to the pulse of a billion dreams.

If I read this even before ChatGPT was a mote in the eye of Karpathy, my eyes would have rolled so far back that metacognitive mindfulness would have become a permanent passive ability stat.

The author of Berserk said it so much better: "Looking from up here, it's as if each flame were a small dream, for each person. They look like a bonfire of dreams, don't they? But, there's not flame for me here. I'm just a temporary visitor, taking comfort from the flame."

>reads like so much pretentious claptrap

>my eyes would have rolled so far back that metacognitive mindfulness would have become a permanent passive ability stat

Yes, I agree it is impressively lifelike, just like it was written by a real flesh-and-blood New Yorker.

I think it reads like an ad for what people think a real New Yorker sounds like. No one’s sticking around for you to ramble about the social weather.

I lived in NYC for a few years, and there are definitely lots of New Yorkers who would find themselves described in the quoted passage.

Or it can be a piece from the New Yorker

Right. All of that stuff is published/performative fluff, not how people actually think or act.

Feeling implicated?

The city is much more romanticized than you think.

Sounds to me like a description of what it feels like to live in any sort of metropolis. Georg Simmel, one of the first sociologists, wrote similar observations on how city dwellers (in 1900s Berlin) would mentally adapt themselves within the context of this city in his 1903 work »The Metropolis and Mental Life«.

The anonymity and (from the POV of a country pumpkin: rudeness) is basically a form of self-protection, because a person would go mad if they had to closely interact with each human being they met. Of course this leads to its own follow-up problems of solitude within a mass of people etc.

But if you e.g. come from a close-knit rural community with strict (religious?) rules, a typical rural chit-chat and surveillance, simething like the anonymity of a city can indeed feel like a revalation, especially if you see yourself as an outcast within the rural community, e.g. because you differ from the expected norms there in some way.

I am not sure about the English translation of Simmels text as I read it in German, but as I remember it it was quite on point, not too long and made observations that are all still valid today, recommended read.

> reads like so much pretentious claptrap

Somewhere out there, there's an LLM that reads exactly like us pretentious HN commentators. All at once cynical, obstinate, and dogmatic. Full of the same "old man yells at cloud" energy that only a jaded engineer could have. After years of scar tissue, overzealous managers, non-technical stakeholders, meddling management, being the dumping ground for other teams, mile-high mountains of Jira tickets.

Techno-curmudgeonism is just one more flavor that the LLMs will replicate with ease.

Different from your comment, that comes from your insight from your lived experiences in this community, the quote is a whole bunch of nothing.

Maybe the nothingness is because LLM can't reason. Maybe it's because it was trained a bit too much in marketing speak and corporate speak.

The thing is, while LLMs can replicate this or that style, they can't actually express opinions. LLM can never not be fluff. Maybe different types of AI can make use of LLM as part of their components, but while we keep putting all of our tokens in LLMs, we will forever get slop.

> LLM can't reason

> we will forever get slop

I think the New York quote is a lot better than whatever I wrote. A human saw that diamond in the rough and decided to share it, and that was the transformative moment.

A human curator of AI outputs, a tastemaker if you will, can ensure we don't see the sloppy outputs. Only the sublime ones that exceed expectations.

I suspect that artists and writers and programmers will pick and choose the best outputs to use and to share. And then they do, that their AI-assisted work will stand above our current expectations of whatever limitations we think AI has.

Eh, I'll take it any day over the cocaine-fuled technooptimism

I see this "I can detect AI bcuz its cringe" view everywhere but all the double blind tests show that people like AI poetry and AI writing slightly more than the greats. Also the other comment is spot on in that this reads exactly like some op-ed articles in nyt's lol.

Domain experts seem to routinely do better at picking out slop, but I'd like an actual study. I was unimpressed with the AI art turing test.

Or a Ghost in the Shell perspective.

I'd say it reads like propaganda for a place i do not live.

Sounds great to the target audience but everyone else just rolls their eyes.

Yes. I realize some people like places like that. You couldn’t pay me to live there.

Reply on news.ycombinator.com