Discussion
AI and bots have officially taken over the internet, report finds
jruohonen: Well, IoT traffic peaked "human traffic" already long ago, Netflix etc. eat a lot of bandwidth, etc., so I am not sure where the news is exactly.
theshrike79: [delayed]
minsung0830: The bigger concern is what happens when AI models start training on AI-generated content at scale. We're already seeing model collapse in research papers where output quality degrades when training data is contaminated with synthetic text. The internet becoming majority bot content basically guarantees this becomes a real problem for the next generation of models.
jve: > This notion of machine bad, human good just is not realisticGlad I found this quote. It is quite helpful for an AI to search the web on behaolf of me... even if it was finding where I can buy particular/similar peanuts locally I got from abroad.
tokioyoyo: Content providers will not agree with this decision, because machine browsing = no ads. Until that gets resolved, I don’t see incentives to align, since any free search requires ads for continuous business.
teleforce: There're human-to-human (H2H), human-to-machine (H2M) or vice versa, and machine-to-machine (M2M) data communication.If you perform simple extrapolation, the M2M data only surpass the others around 2029.Coincidently, in the original timeline of Transformer movie, 2029 is the year that the Resistance, led by John Connor, destroyed Skynet and ended the war against the machines.
cooloo: That's very interesting question I'm ponder about. If all content is AI generated where innovation will come from? Maybe we should differentiates AI assisted content from AI garbage content.
kakacik: Are you saying that high quality human-curated content will be rare and more appreciated in the future compared to endless cheap slop? Can't say I am sad, in contrary
minsung0830: Exactly. The value of verified, human-sourced content goes up as synthetic noise increases. It's basically supply and demand. We might end up in a world where provenance matters more than the content itself
nareyko: One interesting dynamic here is that AI increases content supply much faster than human attention grows.Which means filtering and ranking systems become the main bottleneck.That pushes platforms toward stronger algorithmic selection and sometimes stronger convergence of attention.
i5heu: I think that >in the future< this will be a non problem as there is reality itself that is a much better validator for behavior then human text.We already see this with synthetic training data that basically uses logic in form of math and code as constraint.
SanjayMehta: It's hyperbole, click bait.
SanjayMehta: "Officially?"Who is this official making this pronouncement?
slopinthebag: At the end of the day, the value of producing content will drop to zero and the value of curation content will skyrocket.
b65e8bee43c2ed0: social media was raw sewage even before AI, and WWW was 90% SEO spam generated by third worlders for $2 a day.
axegon_: It already is a problem and maybe unpopular opinion but... That's a good thing. The LLM collapse can't possibly come soon enough. In principle, LLMs can be a good thing but they can't overcome human nature - laziness and the unstoppable desire to take the shortest path. It's those two things that have turned the internet into the absolute dump it is today. Not to mention the bullshitter economy, as I like to call it and everything that comes with it. And all things considered, society does need some reset at this point, the AI bubble might be a good place to kick things off.
senectus1: I wonder if they will start deliberately not scraping social media because of the low quality human content and AI sloppyness of it.suddenly the confirmed quality of the scraped data will be at a premium.. "Scrape Engine Optimizers" ?
eru: > The internet becoming majority bot content basically guarantees this becomes a real problem for the next generation of models.Only if you assume that people who train models are stupid.
RealityVoid: It could be serving ads if they could persuade the machines to do the purchase.
kdheiwns: Stupidity has nothing to do with it. AI articles and comments are now posted everywhere and presented as human. It's becoming harder and harder to determine whether text was written by humans or AI. Where are they supposed to find content to train AI on that isn't polluted with AI content that'll result in a feedback loop? It's like trying to get pure soil and water for growing that's not contaminated with microplastics/nanoplastics and PFAS. There was a time where it was possible. Not anymore. The filth is everywhere and impossible to filter.
thinkingemote: [delayed]
troupo: > Only if you assume that people who train models are stupidSomeone in the chain will be. Even the smartest people buy a lot of their training datasets. What happens when those get contaminated?
raincole: This notion isn't just unrealistic, but extremely dangerous. If we accept "machine bad, human good" line of thinking, the only logical conclusion is that we'll have to verify our biometric every time we'd like to access the internet. Like the UK age verification but 100x worse.
Cthulhu_: If you can devise a tool that can detect AI generated content, you can use it to filter data. But the harsh truth is that "gold standard" training data is from before 2022 or whenever the cutoff was.And even that needs to be curated because before AI tools there was bot content filling up the internet....and even without bots, a lot of human authored content are low value, poorly written, etc.There are (probably) companies out there whose business is to create, curate and improve training sets.
endymion-light: I view a lot of the AI/Bot internet to be slightly a false misnomer. Even before ChatGPT, the degredation of online content was already happening - SEO farms, worsening google search. Most articles you'd find online would be paywalled, most information about specific things would turn out to be a frustrating SEO labyrinth.The current one is awful, and there's so much AI/Bot content, but I can find far more detailed information using AI enabled search that isn't covered in ads. I can get an initial overview of methodology without trawling through SEO articles.I think AI has been almost a natural response to the enshittification of the internet - ChatGPT wouldn't seem so transformative if google search was working like google search rather than ad generator 5000 before it released.
Cthulhu_: Yeah, the internet has been shitty for uh. decades now. 15 odd years ago people were already complaining about listicles and youtube comments.Best thing to do is to avoid idly browsing social media and curate your internet experience.
asdff: Why should they care about new content? Game over already. Just keep regurgitating the same slop to the masses. Even before ai it was like this. How many 2 minute pop songs use the same chord structure? Just keep selling the same thing slightly permutated (or not) from the last. That's capitalism, baby. This isn't a science.
nicbou: A lot of people were running original websites, reviewing stuff, blazing new trails, making new art.It's just harder when you cut all traffic to them, devalue their work and fill the air with AI noise.
woctordho: It's not a problem at all. Humans also read books written by humans.
nicbou: It will not collapse, just be there and disappoint us.
nicbou: Will AI test recipes and try to find the cafe with the nicest vibe? Will it do original research about things that have never been written down yet?I've heard this argument before, but you don't need to think too hard to see the limitations of a machine with no senses.
kdheiwns: And if there's a way to detect that content is AI generated, then there's demand to generate content that seems more human. And we're already at a point where most people have been tricked into believing AI content was real at some point and never even realized it. It'll only get worse.Probably the only real way to validate content is real is building a validation system into devices. Confirm when a photo is taken and send an ID to a server, then when photos are shared, its ID is compared to the image on the camera/phone manufacturer's server. For text, validate every little key press. And there are still ways to game these systems, but I would not be surprised if they're introduced to mitigate AI diffusing everywhere.
Peritract: Not stupid, but I think it's fair to say "careless about/unaware of the wider impact of their work".
asdff: Very few of them actually made money though, compared to people who tried to just take an existing idea they could already order from china in bulk and market the living hell out of it. These companies obviously don't care about art and stuff like that, they really just care about the money.
nicbou: The value to consumers goes up, but that's pointless if they are drowned out by AI overviews paraphrasing their work, half a page of sponsored results, then AI-written SEO spam.I'm a creator of such content, and like everyone else, I have to make do with 60-70% less traffic now.
