Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Three sites made 215,128 “best software” pages for AI. Perplexity cites them

trellner.com

Post reply on HN
Print view View original

231–240 of 265 posts

Re: Three sites made 215,128 “best software” pages for AI. Perplexity cites them

#231

Post by sodapopcan »

eddythompson80 wrote: post #174
sodapopcan wrote: post #96

Earlier quoted context omitted.

SEO is what ruined the web AFAIC. > Time to start some human-only darknets. I know very little about darknets. How could you ensure that they are human-only?

The only way to generate a human-only net is to figure out how to remove all economical incentives from there. The “golden age of the internet” a lot of people, including myself, are nostalgic for had very very little economical incentives. I missed usenet, but I remember being active on small forums in the late 90s and early 2000s that were almost entirely passion projects. I recall a huge controversy in one forum w…

Ya, "how do we make money off this?" ruins everything, but it's inevitable. But yes, that is exactly what I liked about the internet back then: I was just browsing the crazy shit people would put online just because they thought it was cool. I'd still have to go to the store to buy stuff and see friends/people in person. I've started doing those things more again.

Re: Three sites made 215,128 “best software” pages for AI. Perplexity cites them

#232

Post by paulddraper »

xpct wrote: post #18

If I recall correctly, there were some papers which suggested that LLMs favor LLM-generated passages over human written ones. I can consistently reproduce this by asking Claude which code snippet it prefers: the one it generated in a different chat, or one that I refactored for my own needs and find more useful. It always picks its own :) I've also experienced that both Claude and Codex routinely include generated we…

Spoiler: developers also prefer their own code over others’.

Re: Three sites made 215,128 “best software” pages for AI. Perplexity cites them

#233

Post by vetler »

thephyber wrote: post #133
mstaoru wrote: post #47

Well it's not only this, or protection from LLMs training on LLM output. LLMs training on human output is also problematic. I was traveling to an obscure small town, doing some "research" with LLMs beforehand. Every and each one told me enthusiastically to go to "Foobar square" (name changed) for the "best street food in XYZ town", some added a lot of colorful details. There was no Foobar square in XYZ town. There wa…

Your one example doesn't make all of LLMs a lie. It's like someone reading a National Enquirer article about "Hillary Clinton being an alien from outer space" (a real headline topic from decades ago) and drawling the conclusion that all journalism is "a lie". The user has to understand media literacy and be at least a little skeptical of the claims that are made, then cross reference with another source.

> Your one example doesn't make all of LLMs a lie.

It's not a lie though, because the truth isn't guaranteed by the mechanism that generates the answer.

Re: Three sites made 215,128 “best software” pages for AI. Perplexity cites them

#234

Post by ChoGGi »

mschuster91 wrote: post #200
spockz wrote: post #196

Earlier quoted context omitted.

Why is this significant? Many sites are setup like that.

Because the pro-Palestine folks just love to drag their pet peeve into literally everything [1]. [1] https://techforpalestine.org/vercel

Not liking companies that support genocide is a pet peeve?

Re: Three sites made 215,128 “best software” pages for AI. Perplexity cites them

#235

Post by ljf »

Interesting that this remains on the front page @dang - this and the other 3 sites the poster submitted (two others yesterday and one 12 days ago) all appear to follow the same AI generated pattern, all newly registered, AI written and with vague 'about us' pages. I'm not sure if Jakob Greenfeld registered/owns them all, or if it is linked to his marketing/sales business - but it is rather fishy.

Re: Three sites made 215,128 “best software” pages for AI. Perplexity cites them

#236

Post by eru »

kqr wrote: post #171
xpct wrote: post #18

If I recall correctly, there were some papers which suggested that LLMs favor LLM-generated passages over human written ones. I can consistently reproduce this by asking Claude which code snippet it prefers: the one it generated in a different chat, or one that I refactored for my own needs and find more useful. It always picks its own :) I've also experienced that both Claude and Codex routinely include generated we…

I noticed this when I tried to get LLMs to play text adventures. Early on in my experiments, I wanted to give them hints when they got stuck at a puzzle. I did this by stopping the loop, injecting thoughts into the LLMs own persistent scratchpad (as if it had thought of that itself), and then starting the loop up again.[1] That way, the LLM would read what it had intended to remember from the previous turn including…

> I speculate this is part of how they can be useful in the first place, despite their non-determinism.

Humans and monte carlo simulations are also non-deterministic and can be useful. So I don't see much of a need to explain why non-deterministic system can be useful.

Re: Three sites made 215,128 “best software” pages for AI. Perplexity cites them

#237

Post by eru »

joquarky wrote: post #146
grumbelbart2 wrote: post #88

Earlier quoted context omitted.

Sure maybe, but the online ad market is USD 400b give or take, which vastly outmatches any such budgets.

Imagine what humanity could accomplish if marketing were taboo.

How are you going to learn about useful new goods and services?

It is already legal for companies not to do any marketing.

Re: Three sites made 215,128 “best software” pages for AI. Perplexity cites them

#238

Post by nulbyte »

fluoridation wrote: post #143
thephyber wrote: post #133

Earlier quoted context omitted.

Your one example doesn't make all of LLMs a lie. It's like someone reading a National Enquirer article about "Hillary Clinton being an alien from outer space" (a real headline topic from decades ago) and drawling the conclusion that all journalism is "a lie". The user has to understand media literacy and be at least a little skeptical of the claims that are made, then cross reference with another source.

Okay, but the anecdote states that every model repeated the pseudo-factoid about Foobar square, not just the 4 GB open source model equivalent of a tabloid.

I think the key phrase here is, "an obscure small town." There may only be a single mention of this place, hence the only one on which a response can be based. This says more about the user's understanding of LLMs than it does about LLMs.

Re: Three sites made 215,128 “best software” pages for AI. Perplexity cites them

#239

Post by latexr »

ljf wrote: post #235

Interesting that this remains on the front page @dang - this and the other 3 sites the poster submitted (two others yesterday and one 12 days ago) all appear to follow the same AI generated pattern, all newly registered, AI written and with vague 'about us' pages. I'm not sure if Jakob Greenfeld registered/owns them all, or if it is linked to his marketing/sales business - but it is rather fishy.

@mentions aren’t a thing on HN. If you want to contact Dan and Tom, use the “Contact” at the bottom of the page. They are very responsive.

Re: Three sites made 215,128 “best software” pages for AI. Perplexity cites them

#240

Post by eru »

greggoB wrote: post #168
marcus_holmes wrote: post #151

Earlier quoted context omitted.

Does an LLM have any idea of what "best" is? I think it doesn't, and just predicts the range of most statistically likely next tokens based on its training data, and picks one of those.

In ML, "best" is what minimises the loss function, so in a certain (very basic) sense it does.

You are mixing up levels.

When you are asking it a question (like which of these two texts is the best), the output is also just picked by minimising that loss function. There's no guarantee that answering "Text B is better" aligns with text B minimising the loss function.

(And they aren't really minimising loss functions during inference. They sample from a distribution. During training they minimise the loss function of the distribution.)

Post reply on HN