Over the past few months, various academics and AI companies have attempted to predict how artificial intelligence will impact the labor market. These studies, including a high-profile paper published by Anthropic earlier this month, primarily focus on what AI is good at, or could be good at, and try to fit it into existing jobs and roles. But these papers ignore some of the most influential and most common uses of AI today: AI porn and AI slop.
Anthropic paper “”AI’s impact on the labor market: New measures and early evidence“” is basically trying to find a one-to-one correlation between the tasks people are doing today at work and what people are using Claude for. Researchers are also trying to predict whether a job task is “theoretically possible with AI.” As a result, this graph was created, which became a bit of a hot topic, Published in Newsletter by MSNOW’s Phillip Bump and Article by technology journalist Christopher Mims. (Because everything is terriblethis research is now also available on gambling websites where you can see the clear probability of your job being replaced by an AI. )
In the thread, Mims claims that the “theoretical ability” of AI to do different tasks in different fields is a complete fabrication, and that this graph basically means nothing. Mr. Mims makes a good and fair observation. The nature of so many studies that attempt to predict which people will lose their jobs to AI is flawed. This is because the input requires some guessing.
However, I believe that most of these studies are flawed in a deeper sense. Anthropic fails to consider how people actually use AI in real life, even though it claims that’s exactly what it actually does. “We introduce Observed Exposure, a new measure of AI displacement risk that combines theoretical LLM capabilities with real-world usage data and weights automated (as opposed to augmentative) and work-related use more heavily,” the researchers wrote. This is partially based on “.human economy index” was. Featured in a very long paper published in January It attempts to catalog all the noble uses of AI in specific work-related situations. These uses include “completion of academic assignments in the humanities and social sciences across multiple disciplines,” “drafting and revising professional and business communications in the workplace,” and “building, debugging, and customizing web applications and websites.”
Very common uses of AI, such as “creating AI porn” and “creating AI slop or spam,” are not included in Anthropic’s research. These uses destroy discoverability on the internet and cause cascading social and economic harm. Researchers seem too picky or too embarrassed to address the fact that people like to use AI to make porn, and people like to use AI to spam social media and the internet, essentially causing economic harm to creators, adult performers, journalists, musicians, writers, artists, website owners, small businesses, etc. Written in the first 404 Media Generative AI Market AnalysisPeople Love to Cum, and many of the most popular generative AI websites are explicitly focused on creating AI porn and non-consensual AI porn. Anthropic’s research continues a long tradition of AI companies wanting to emphasize the “good” uses of AI that appear in their marketing materials while ignoring the world-shattering uses that people actually use AI for. (While it may be true that people are using Claude at a disproportionately high rate for more traditional work applications, research on “AI’s labor market impact” should not focus on the use of a single tool and extrapolate it to all other tools. Claude’s jailbreak version Very popular among sexbot enthusiasts).
Meanwhile, as we have repeatedly shown, social media websites and Google search results are largely being replaced by advances in AI. Chatbots themselves have killed traffic to many websites that once could rely on ad revenue to hire people.
The Anthropic paper attempts to estimate how AI will impact “arts and media,” but again, the way the researchers do this is by trying to determine whether AI can directly perform the tasks that AI researchers envision being required of people in “arts and media” jobs. Other widely cited papers on AI-related unemployment also don’t really try to consider the potential macro effects of ongoing internet dysfunction and zombification, instead focusing on “AI exposure.” This is primarily an attempt to predict or measure whether AI or LLM can directly replace certain tasks that need to be performed by humans. widely cited papers National Economic Research Bureau and brookings Studies released over the past few months attempt to determine the readiness of workers in certain sectors to have many tasks automated by AI. The Brookings paper at least mentions the possibility of unpredictable society-wide change: “The evidence underlying the estimation of adaptive capacity here comes primarily from the observed effects of localized displacement events, rather than from large-scale employment movements across occupations. As a result, this index… may be most beneficial when migration is relatively isolated, for example when workers lose their jobs but related occupations remain stable. In scenarios where AI affects clusters of related occupations simultaneously, structural job availability may become important.” If we fundamentally transform the economy on a scale comparable to the Industrial Revolution (as some experts suggest is possible), entire skill sets could overlap across multiple occupations simultaneously. ”
Let me be clear: AI-induced unemployment is an important thing to study and consider. But so many jobs, side hustles, and economic activities rely more broadly on the Internet, or social media in its broadest sense. According to study after study, Google is getting worse and worse, Traffic to your website is decreasingand the amount of both is increasing Web traffic and web content is generated by AI and bots. Anecdotally, creators and influencers are telling us: It becomes harder to compete with AI slop And it’s hard to justify spending days or weeks creating content just to publish it to a platform where your AI competitors can effortlessly bypass recommendation algorithms. I heard from a website that had to do it. lay off employees or close down Either because Google’s AI overview destroyed your web traffic or because you lost out in search engine rankings due to AI degradation. Authors regularly compete AI publishes plagiarized versions of books on Amazonand Spotify is flooded with AI-generated musictoo.
All of this shows that these studies on the economic impact of AI ignore the very important context that AI is eating away at and disrupting the internet and social media. We are moving from the many-to-many publishing environment that created countless jobs and businesses to a system where AI tools can easily overwhelm human-created websites, businesses, art, writing, video, and human activity on the internet. What is happening may be too chaotic, confusing, and unpleasant for AI companies to ignore, but it would be illegal to ignore it completely.
