Are We Watching The Internet Die?
We recorded 29 claims from this post. 11 were checked against evidence. 18 can’t be checked. 7 more are held. Read the post at its source.
Claims from this post (29)
In the order they appear in the post. Each quotation is the speaker’s words in this post. A label such as “Supported” or “Can’t be checked” is the claim’s current assessment; its page has the evidence and the reasons.
Assessment: Can’t be checkedSaid
“Yet what's happening to the web is far more sinister than simple greed, but the destruction of the user-generated internet, where executives think they've found a way to replace human beings making cool things with generative monstrosities trained on datasets controlled and monetized by trillion-dollar firms.”
What we tested: What is happening to the web is the destruction of the user-generated internet, in which executives think they have found a way to replace humans making things with generative AI output trained on datasets that trillion-dollar firms control and monetize.
Assessment: SupportedSaid
“The algorithms are easily-tricked, and the tools used to trick them are becoming easier to use and scale.”
What we tested: Platform algorithms are easily tricked, and the tools used to trick them, including generative AI, are becoming easier to use and scale.
What happened: The report describes using Byword to generate 1,800 articles in hours and obtain search traffic, supporting accessible production at scale. It also reports that much of the traffic subsequently disappeared; figures are estimates and campaign claims, not audited platform records. (The Decoder, )
Assessment: Can’t be checkedSaid
“And it's slowly killing the internet.”
What we tested: Generative content that games algorithms is slowly killing the internet.
Assessment: UnresolvedSaid
“As more internet content is created, either partially or entirely through generative AI, the models themselves will find themselves increasingly inbred, training themselves on content written by their own models which are, on some level, permanently locked in 2023, before the advent of a tool that is specifically intended to replace content created by human beings.”
What we tested: As more internet content is made with generative AI, models will be trained more and more on AI-generated content and become increasingly inbred.
Assessment: UnresolvedSaid
“In reality, a Habsburg AI will be one that is increasingly more generic and empty, normalized into a slop of anodyne business-speak as its models are trained on increasingly-identical content.”
What we tested: A Habsburg AI, meaning a system heavily trained on the outputs of other generative AIs, will become increasingly generic and empty, normalized into anodyne business-speak, as its models are trained on increasingly identical content.
Assessment: Can’t be checkedSaid
“While Google has claimed that their upcoming "core" update will help promote "content for people and not to rank in search engines," it’s made this promise before, and I severely doubt anything meaningfully changes.”
What we tested: The author severely doubts that anything meaningfully changes after Google's upcoming core update, which Google claims will promote content for people rather than content made to rank in search engines.
Assessment: UnresolvedSaid
“And because these platforms were built to reward scale and volume far more often than quality, AI naturally rewards those who can find the spammiest ways to manipulate the algorithm.”
What we tested: Because platforms reward scale and volume over quality, generative AI rewards people who find the spammiest ways to manipulate algorithms.
Assessment: Can’t be checkedSaid
“This hasn't stopped spammers from publishing awkward rewrites and summaries of other people's books, and because Amazon's policies don't outright ban AI-generated content, ChatGPT has become an inoperable cancer on the body of the publishing industry.”
What we tested: Because Amazon's policies don't outright ban AI-generated content, ChatGPT has become an 'inoperable cancer on the body of the publishing industry.'
Assessment: Can’t be checkedSaid
“These stories are, of course, all manifestations of a singular problem: that generative artificial intelligence is poison for an internet dependent on algorithms.”
What we tested: The stories described are all manifestations of a single problem: generative AI is 'poison' for an internet dependent on algorithms.
Assessment: SupportedSaid
“Generative AI is a perfect tool for soullessly churning out content to match a particular set of instructions — such as those that an algorithm follows — and while an algorithm can theoretically be tuned to evaluate content as "human," so can scaled content be tweaked to make it seem more human.”
What we tested: Generative AI is a perfect tool for soullessly churning out content to match a set of instructions such as an algorithm's, and while an algorithm can theoretically be tuned to evaluate content as human, scaled content can likewise be tweaked to seem more human.
What happened: The researchers generated tens of thousands of instruction-following examples, establishing repeatable production at scale. Their accompanying evaluation and limitations do not establish flawless adherence or universal content quality. (Stanford Center for Research on Foundation Models, )
Assessment: Can’t be checkedSaid
“Google might pretend it cares about the quality of search results, but nothing about search's decade-long decline has suggested it’s actually going to do anything.”
What we tested: Google might pretend it cares about search result quality, but nothing about search's decade-long decline has suggested Google is actually going to do anything.
Assessment: Can’t be checkedSaid
“As we speak, the battle that platforms are fighting is against generative spam, a cartoonish and obvious threat of outright nonsense, meaningless chum that can and should (and likely will) be stopped.”
What we tested: The battle platforms are currently fighting is against generative spam, obvious outright nonsense that can and should, and likely will, be stopped.
Assessment: UnresolvedSaid
“Like a plant turning to capture sunlight, parts of the internet have already twisted toward the satisfaction of algorithms, and as others become dependent on generative AI (like Quora, which now promotes ChatGPT-generated answers at the top of results), so will the web become more dependent and dictated by automated systems.”
What we tested: As more sites come to depend on generative AI, the web will become more dependent on, and more dictated by, automated systems.
Assessment: UnresolvedSaid
“The ultimate problem is that this morass of uselessness will lead companies like Google to force their generative AIs to "fix" the problem by generating answers to sift through the crap.”
What we tested: Low-quality content will lead companies like Google to use generative AI to produce answers that sift through it.
Assessment: Can’t be checkedSaid
“There could be a future where the majority of web users experience the web through a series of portals, like Arc Search's "browse for me" feature, which visits websites for you and summarizes their information using AI.”
What we tested: Most web users could come to experience the web through AI summarization portals.
Assessment: SupportedSaid
“Yet at some point these portals will generate an answer based off of the data they have (or have access to, in the case of Tumblr and Wordpress) rather than linking you to a place where you can find an answer by reading something created by another person.”
What we tested: At some point search portals will give AI-generated answers from their data instead of linking users to content made by people.
What happened: Google announced public US rollout of AI Mode, describing retrieval across web subtopics and an in-search response experience. The same announcement retains source links, which limits support to direct answers substituting for necessary source visits. (Google, )
Assessment: Can’t be checkedSaid
“Their eventual goal, in my mind, is to remove as much interaction with the larger internet as possible, summarizing and regurgitating as much as they can so that they can control and monetize the results as much as possible.”
What we tested: In his view, the major platforms aim to minimize interaction with the wider internet by summarizing content with AI so they can control and monetize the results.
Assessment: Can’t be checkedSaid
“As a result, they can't be too picky when it comes to the information they choose, meaning that they're more than likely going to depend on openly-available content from the internet, which as I've suggested earlier will become increasingly normalized by the demands of algorithms and the ease of automating the generic content that satisfies them.”
What we tested: OpenAI's and Anthropic's models will more than likely depend on openly available web content, which will become increasingly normalized and generic.
Assessment: UnresolvedSaid
“OpenAI's ChatGPT and Anthropic's Claude are dependent on a constant flow of training data to improve their models, to the point that it's effectively impossible for them to operate without violating copyright.”
What we tested: OpenAI's ChatGPT and Anthropic's Claude depend on a constant flow of training data to improve their models, to the point that it is effectively impossible for them to operate without violating copyright.
Assessment: UnresolvedSaid
“This is why so many models' responses sound so similar — they're heavily dependent on the stuff they're fed for their outputs, and so much of their "intelligence" comes from the same training data.”
What we tested: Many models' responses sound alike because they depend heavily on the same training data.
Assessment: Can’t be checkedSaid
“It's a different flavor of the same problem — these models don't really "know" anything. They're copying other people's homework.”
What we tested: These models don't really 'know' anything; they're 'copying other people's homework.'
Assessment: Can’t be checkedSaid
“This may be why Sam Altman is trying to claim that giant AI models are not the future — because there may not be enough fuel to grow them much further.”
What we tested: There may not be enough training data to grow giant AI models much further, which may be why Sam Altman says they are not the future.
Assessment: Can’t be checkedSaid
“ChatGPT's DALL-E (images) and Sora (video) products are, as I've noted, limited by the availability of ways to teach them as well as the limits of generative AI itself, meaning that video may continue to dominate the internet as text-based content finds itself crowded out by AI-generated content.”
What we tested: OpenAI's DALL-E and Sora are limited by the availability of ways to teach them and by the limits of generative AI itself, meaning video may continue to dominate the internet as text-based content is crowded out by AI-generated content.
Assessment: Can’t be checkedSaid
“There's also no way to escape the fact that these hungry robots require legal plagiarism, and any number of copyright assaults could massively slow their progress.”
What we tested: Generative AI models require 'legal plagiarism', and any number of copyright assaults could massively slow their progress.
Assessment: Can’t be checkedSaid
“It's incredibly difficult to make a model forget information, meaning that there may, at some point, be steps back in the development of models if datasets have to be reverted to previous versions with copyrighted materials removed.”
What we tested: Because models are hard to make forget information, model development may go backwards if datasets have to be reverted to remove copyrighted material.
Assessment: Can’t be checkedSaid
“The numerous lawsuits against OpenAI could break the back of the company, and while Altman and other AI fantasists may pretend that these models are an intractable path to the future of society, any force that controls (or makes them pay for) the data that they use will kneecap the company and force them to come up with a way to make these models ethically.”
What we tested: The many copyright lawsuits against OpenAI could break the company.
Assessment: UnresolvedSaid
“The numerous lawsuits against OpenAI could break the back of the company, and while Altman and other AI fantasists may pretend that these models are an intractable path to the future of society, any force that controls (or makes them pay for) the data that they use will kneecap the company and force them to come up with a way to make these models ethically.”
What we tested: If any force controls OpenAI's training data or makes it pay for that data, OpenAI will be kneecapped and forced to find an ethical way to build models.
Assessment: Can’t be checkedSaid
“Those who rely on generative AI to create are not creators any more than a person that commissions a portrait is an artist.”
What we tested: Those who rely on generative AI to create are not creators, any more than a person who commissions a portrait is an artist.
Assessment: Can’t be checkedSaid
“Don't accept that AI "might one day" be great. Demand that it is today, and reject anything less than perfection from men that make billions of dollars shipping you half-finished shit.”
What we tested: People should not accept that AI 'might one day' be great, but should demand that it is great today and reject anything less than perfection from the men making billions shipping half-finished products.
Held from this post (7)
These records did not pass every check, so nothing about them is published. Held is not a finding.
- The two reviews read the claim differently
- The record does not meet the current rules. Our reading, not assessed: Generative AI tools are fancy autocomplete that can't avoid repeating mistakes learned from stolen training data.
- The two reviews read the claim differently
- The two reviews read the claim differently
- The two reviews read the claim differently
- The two reviews read the claim differently
- The two reviews read the claim differently