AI garbage is already polluting the internet | Hacker News Reader

AI garbage is already polluting the internet

nytimes.com

45 pointsfurcyd26 comments

I disagree with the authors assertion that a simple watermark would solve the issue. I don't see how you can ever enforce something like a "Clean Air act" against AI. AI is a spectrum and thus it will be a gray area.

It would require a draconian system in place that would make people swear that content their posting isn't AI and then have steep penalties for breaking "the oath". But what if someone used a spell checker? Grammar checker? or sentence rewriting in Grammarly? should they be punished for breaking the "clean internet act"?

If an honor system doesn't work, then one possibility is to use an economic system so that there is a cost and value associated with the content being produced. However as we have seen with the decline of journalism, is that is hard to get people to pay for quality content. so the cheaper content would still win.

Maybe something like a legal tax for content produced could work though. If you produce over Xmb of content as an individual you start to get taxed on your "pollution" to the internet. Single internet user just posting: no tax, influencer posting youtube videos: small tax, AI company dumping out 1000s of books and 1000s of videos: big tax.

The size difference between text and video is enough to kill a tax per gigabyte proposal.

>tax

Don’t see why politicians should get to profit from it

Do you just not understand what a tax is?

It is naive to assume increased tax revenue doesn’t equate to increased power for those who control how that money is spent.

It is OK to make an argument against taking power from one group by force and giving to another.

Okay sure, we can make that argument. But that's not what they said. "Profiting" off public money is just embezzlement.

Democratic power puts it in the hands of people. Now, if you’re saying that your democracy doesn’t serve its people — I wouldn’t blame the politicians. You might want to take a look at why politicians don’t act in the interests of the population

inb4 "This article contains text known in the state of California to be processed by AI"

Analogy is to plagiarism. The normal person content industry seems to do an okay job ejecting plagiarists and they should be encouraged to do the same with ai generated sludge.

It doesn't need to be 100% perfect or watertight. Almost no law is.

All laws are easy to break. I could quite literally walk up to someone and punch them in the face right now. And chances are I won't be caught.

I could also push someone. Would that be assault? What if I merely touch them? Almost everything is a spectrum and has gray areas.

Original title (though I'll allow for some A/B testing):

A.I.-Generated Garbage Is Polluting Our Culture

And here's a gift link: https://www.nytimes.com/2024/03/29/opinion/ai-internet-x-you...

I appreciate the gift link

The scale is going to keep getting worse though there has been SEO crap, blogspam etc etc since the internet started. This is just the latest. There's an opportunity to build a search mechanism that is immune to the broadening underbelly of the internet, just like there was in the early days.

and how exactly do you propose that this search works?

I assume it will be with a PhD thesis similar to how Google started and not an off the cuff response to an Internet comment.

Although it would be nice if it truly were that easy.

It is that easy, the thesis is already on the internet. reboot pagerank and dump the NLP. Flag sites that create content too fast and let users find webpages that match their query.

This would be difficult today because this would still be a niche thing as that ship has sailed since Google has tethered itself everywhere.

Every time you enumerate another method for making a functional search engine right now, you also posit a way for it to be circumvented in an hour or two.

"Oh, it's easy! Just bring PageRank back, and flag sites that create content too fast!"

Later that afternoon: sites multiply, and individually generate content slowly

It does feel like people don't give Google enough credit that, even in it's current state, they still do a pretty good job in the face of basically every bad actor on the internet.

the pragmatic solution is simply to build your own, personal, dead internet.

Google is at an advantage here. They can see into their own blackbox and instantly change the rules whenever they want. The abusers have to wait a while to see the results of their experiments, knowing that Google can put all that effort to waste instantly.

Pagerank alone was subverted pretty early on, which is why Google started bring additional signals into things.

This article implies most content on the internet wasn’t already pollution anyways. It was already possible to min max SEO with bash scripts, AI doesn’t change any of that.

The internet pollution is being done by humans. They're only now using AI to do so.

> In other words, significant numbers of researchers at A.I. conferences were caught handing their peer review of others’ work over to A.I. — or, at minimum, writing them with lots of A.I. assistance.

AI is the lazy tool at the moment, but the internet was already polluted by humans using other means for other reasons besides publishing papers.

Every technological convenience or economic cost reduction has similar effects: email spam, SMS, automated calls, ...

Wasn't the NY times founded way before the microchip?

zing!

Polluting the internet. lol

Is it really bad, or is it yet another “kids playing games nowadays” bad?

We just invented new instruments for being lazy. The articles were diluted to meet page quota before. The absurd, idiotic videos for kids didn’t start with AI. Models knew about collapse from the start, they’ll figure out.

IMO, the article is a low-key FUD piece with no goal.

Reply on news.ycombinator.com