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

The most talented person in the world

matt.sh

Post reply on HN
Print view View original

121–130 of 155 posts

Re: The most talented person in the world

#121

Post by pona-a »

WithinReason wrote: post #112
Uehreka wrote: post #110

Earlier quoted context omitted.

What makes you think that LLMs will be better at combating spam than they are at creating it? There’s no universal rule that innovations in AI will go hand in hand with innovations in detecting AI, yet I feel like I see people talking all the time like that’s the case. As of right now, LLMs are prolific but unreliable, which makes them extremely well suited for generating spam, but unsuited to detecting it without a…

Classification is easier than generation

Sadly, this didn't seem to track for LLMs. Even OpenAI gave up on trying to detect its own outputs.

> As of July 20, 2023, the AI classifier is no longer available due to its low rate of accuracy. We are working to incorporate feedback and are currently researching more effective provenance techniques for text, and have made a commitment to develop and deploy mechanisms that enable users to understand if audio or visual content is AI-generated.

Re: The most talented person in the world

#122

Post by TeMPOraL »

WithinReason wrote: post #119
Uehreka wrote: post #116

Earlier quoted context omitted.

Yeah no I’m gonna need more than that chief. Everything I know about LLMs says the opposite.

Look up Generative Adversarial Networks, that's their basic principle. You're not classifying just text, you're classifying entire web pages.

I'd say the whole point of GAN is that generation is cheaper than classification, therefore an effective brute-force way of making a good classifier is to generate an infinite supply of examples with a-priori known classification, and pit it against a classifier.

Re: The most talented person in the world

#123

Post by seanhunter »

worstspotgain wrote: post #20
api wrote: post #3

This kind of spam was already destroying the web before LLMs came along. Now it’s being accelerated by thousands of times. Mainstream web search is probably cooked. Kagi and other niche players might have a chance if the fact that they are not beholden to advertisers lets them introduce features to do things like downrank content with ads. Kagi has “small web” which I think includes this in its weighting. Open social…

You're assuming that LLMs will boost spam and scams more than LLMs will tone them down by marginalizing them automatically. I have the opposite view. SMTP spam used to be mostly unavoidable, then one day it was not. IMO, the number of engineers and moderators needed to offset one scammer is about to take a huge dive.

Your analysis overlooks a massive cost asymmetry[1] in favour of the spammer. THey only have to pay the LLM cost once to generate a message which they can use thousands/millions of times versus the receiving side would need to pay an LLM to check and classify each incoming message.

[1]Either you pay a SaaS LLM provider or you pay the cost of compute to run the LLM yourself

Re: The most talented person in the world

#124

Post by TeMPOraL »

parpfish wrote: post #85
jprete wrote: post #79

Earlier quoted context omitted.

I find personal email to be almost useless for anything other than mailing lists and online shopping messages. I don't consider it fixed; we already left it to burn.

I use email constantly but I can’t tell you the last time I saw actual spam in my inbox Sure, there are occasional unwanted marketing emails but those are easily dispatched with unsubscribe and/or inbox filters

Spam evolved. The filters catch most of the egregious kind, but our mailboxes are still flooded with the rest: marketing of legitimate companies, which you likely interacted with at some point, however briefly, and often mixed into transactional messages. This consists the majority of most people's inbox; I bet it's the case for you too.

Re: The most talented person in the world

#125

Post by WithinReason »

pona-a wrote: post #121
WithinReason wrote: post #112

Earlier quoted context omitted.

Classification is easier than generation

Sadly, this didn't seem to track for LLMs. Even OpenAI gave up on trying to detect its own outputs. > As of July 20, 2023, the AI classifier is no longer available due to its low rate of accuracy. We are working to incorporate feedback and are currently researching more effective provenance techniques for text, and have made a commitment to develop and deploy mechanisms that enable users to understand if audio or vis…

You're not classifying just text, you're classifying entire web pages. If it's easy to tell for a human that it's SEO spam, it's easy for a classifier.

Re: The most talented person in the world

#126

Post by ChrisMarshallNY »

illusive4080 wrote: post #63
ChrisMarshallNY wrote: post #21

Earlier quoted context omitted.

> Even those are really having to fight bots though. We recently released a closed-system, iOS-only app, that has a fairly rudimentary, privacy-first system of user registration. It is currently restricted to the US and Canada. Each signup request is manually vetted. There is no automatic registration. The app is designed for a specific demographic, and we do our best to ensure that new accounts are real people, that…

What app and serving whom?

I won’t mention it here.

Like I said, each signup is individually vetted, and the last thing we need, is hundreds of curious geeks, signing up one-shot accounts.

It’s for addicts, seeking Recovery.

We’re not interested in scale; only quality. People’s lives can depend on it.

The bots have declined, recently. I suspect that there’s a watcher bot, that triggers on new apps. When it first came out, we had a lot.

Re: The most talented person in the world

#127

Post by smusamashah »

There was another article here on HN some time ago on how big media publishers are littering the web with spam recommendations https://news.ycombinator.com/item?id=39433451 (https://housefresh.com/david-vs-digital-goliaths/)

This listed review spam from Dotdash Meredith which I used a clue to search and block all sites by them using uBlacklist.

Edit: Found another article saying the same thing there are only few big media giant's dominating search https://detailed.com/google-control/

Re: The most talented person in the world

#128

Post by throwaway290 »

TeMPOraL wrote: post #124
parpfish wrote: post #85

Earlier quoted context omitted.

I use email constantly but I can’t tell you the last time I saw actual spam in my inbox Sure, there are occasional unwanted marketing emails but those are easily dispatched with unsubscribe and/or inbox filters

Spam evolved. The filters catch most of the egregious kind, but our mailboxes are still flooded with the rest: marketing of legitimate companies, which you likely interacted with at some point, however briefly, and often mixed into transactional messages. This consists the majority of most people's inbox; I bet it's the case for you too.

That's not spam. Watch what you subscribe for and periodically unsubscribe, done.

Re: The most talented person in the world

#129

Post by TeMPOraL »

throwaway290 wrote: post #128
TeMPOraL wrote: post #124

Earlier quoted context omitted.

Spam evolved. The filters catch most of the egregious kind, but our mailboxes are still flooded with the rest: marketing of legitimate companies, which you likely interacted with at some point, however briefly, and often mixed into transactional messages. This consists the majority of most people's inbox; I bet it's the case for you too.

That's not spam. Watch what you subscribe for and periodically unsubscribe, done.

I don't subscribe for anything of this kind, and if something somehow gets into my inbox, I report it as spam and then unsubscribe.

All that is very much spam. I'm never interested in marketing material from vendors. Doesn't stop them from padding every single e-mail with it.

Re: The most talented person in the world

#130

Post by boesboes »

sebtron wrote: post #114
fxtentacle wrote: post #108

Earlier quoted context omitted.

As much as I hate the idea, I'm convinced it would be a godsent for 99% of the population. Just imagine when you buy your SIM card the phone shop asks you: Do you want to limit incoming calls to people who you either called before or who have ever had a permanent residence in your country? 99% of spam and scam calls blocked, just like that. And just imagine how hilarious it would be if all those Nigerian prince email…

99% of the population is also happy with carrying a spyware device everywhere and allowing Google to know everything you do in order to serve you better ads. I am not saying this does not have upsides, but it would be a nightmare to have it imposwd on you.

I don't think this is actually true, they are not happy about it, they just can't care because they have no real choice to do it differntly
Post reply on HN