AI Junk Is Starting to Pollute the Internet | Hacker News Reader

AI Junk Is Starting to Pollute the Internet

wsj.com

37 pointssebg17 comments

Content spinning has been polluting the internet for over 15 years, but nobody really noticed. It pains me to think that AI is now being trained on barely coherent junk that was generated to farm Adsense revenue. Junk created by paying someone $1 to write a 800 word article about a topic they know nothing about, then creating a dozen versions of it by replacing verbs and adjectives with synonyms.

If OpenAI et al are actually illegally training on textbooks etc, I say good. At least the information might be somewhat reliable.

People don't get what info pollution has already happened.

Estimate is only 0.5% of content produced is consumed https://www.un.org/sites/un2.un.org/files/attention_economy_...

Naturally platforms don't tell content creators that.

Why is info pollution happening -

1. Everyone is given broadcast capability. Cost to broadcast is 0. Cost to spam is 0. Cost to waste everyone's else's time is 0. A mindless achievement of the platform.

2. Biggest attention craving ppl in society capture the largest chunk of the reward pool. Attention is finite. People have limited attention to give to anything, but many have unlimited capacity to receive it. Such people are the biggest beneficiaries.

3. To combat the pollution that enriches them, the platforms sell a story, that what gets to the top of Search result pages, News feeds, trending pages is the highest Quality content. It's not. Attention capture does not equal quality. And incentive to game the rankings to pseudo signal Quality keeps growing.

Whats the solution? Read the UN report linked above.

I found an obvious one I thought was fairly amusing. This site acting like "The Far Pavilions" miniseries from 1984 had only just come out and was up for a second season next year, complete with a countdown clock - https://whenhbo.com/the-far-pavilions

Slightly off topic but one place where the level of garbage really hits me is looking up recipes. All recipe sites are crammed full of garbage keywords for SEO. If that’s not done by AI it might as well be. I only care about 1) the ingredients 2) cooking instructions and 3) maybe a picture.

If someone would make a wiki recipe that’s the only site I would ever use.

I can’t recall the name but someone will surely reply with what you’re looking for - it does exist but obviously would never show up in search results.

Nothing irritates me more than having to skim a 600 word essay on how this is the perfect meal to make on Wednesdays when Timmy has soccer practice but Sarah needs to be at piano shortly after, and how your Uncle who doesn’t like spices said it was the best meal he’s ever had a potluck where a storm knocked out the power and dishes couldn’t be heated in the oven.

https://en.wikibooks.org/wiki/Cookbook:Table_of_Contents and https://commons.wikimedia.org/wiki/Category:Recipes

https://based.cooking/

That’s the one! Thank you.

As it happens:

https://www.theatlantic.com/technology/archive/2023/07/googl...

I thought maybe fandom has it, and sure: https://recipes.fandom.com/

Modern art of searching became remembering hub sites and avoiding direct-querying actual search engines. Seems we’re almost back to “portals” epoch.

You'll absolutely need that particular Heinz sauce, and himalayan cayenne pepper, best cooked in [producer] cooker. Cooking recipes on the web are dead rotting corpses. The only salvation are ebooks of reputable cooking books.

> I only care about 1) the ingredients 2) cooking instructions and 3) maybe a picture.

Why not buy a proper recipe book then? They offer exactly what you want, with no bait and switch.

You aren't paying for that person's web hosting, what do they care what you want?

Futurama was right. Spammers really are the ultimate evil of the universe.

https://archive.is/vhifn

starting? auto-generated blogspam has been around for at least a decade, it's just worse now.

If a leaf falling into the tank of a porta-potty is pollution, then sure, AI is polluting the Internet.

Depends on your own reading habits. So far I've seen nearly zero AI junk except for the occasional forum post which cleverly ties the previous comment to some corporate link. But that happened like twice this year so I'm not super worried.

If a blogspam blogspams, but no one reads it, is it really a problem?

Obviously I'm excluding deliberately AI generated images, videos, etc. as they tend to bring a genuine entertainment value.

Reply on news.ycombinator.com