The article describes how reputable sites blocking AI crawlers forces AI models to rely on low-quality, AI-generated content, leading to issues like post-hoc citation and infrastructure-level confirmation bias, and offers advice for better source verification.
The internet is inbreeding. Lately, when I ask AI for sources, I barely recognize any of them. Not the "oh, a niche blog I hadn't seen before" kind of unfamiliar — more like websites that seem to exist for the sole purpose of being cited by an AI. So I looked into why that's happening. Turns out most of the reputable, high-quality sites are now blocking AI crawlers entirely. Which means if the good sources are locked out, what's actually left for these models to pull from? Mostly AI-generated rewrites of other AI-generated pages. Content farms built specifically to get quoted by chatbots. And more and more, brands publishing their own "research," where the real purpose is marketing rather than information. There's also a pattern called post-hoc citation, where the model lands on an answer first, then hunts down sources that happen to support it afterward. Honestly, that's not so different from how a lot of us wrote essays back in school. The machines really did learn from the best. But when a student does that, maybe one teacher notices and gets annoyed. When it's the default research tool for a billion people, that same habit turns into infrastructure-level confirmation bias. So what's actually worth doing about it? Ask the AI to also pull sources that contradict its own answer. Open the citations yourself. If a site wouldn't earn your trust coming from a person, it shouldn't earn it coming from a model either. Trace a number back to where it originally came from, not just whoever last repeated it. Most cited "sources" are several steps removed from the original. We were sold "trained on the internet." What we're actually getting looks a lot more like "trained on the internet's marketing department.
AI is causing online content to be optimized for machines, leading to data scraping and a decline in original website traffic, replacing human content with synthetic alternatives.
The article questions whether the proliferation of AI-generated content on the internet could diminish the quality of human-created information and hinder future AI learning by creating a feedback loop of AI training on AI-generated data.
The article raises concerns about the long-term impact of AI-generated content polluting the internet, making it difficult to verify authenticity and grounding in reality, with severe consequences for future AI-governed systems.
The article argues that the proliferation of low-quality, AI-generated content ('AI slop') on platforms like GitHub and blogs is degrading the value of online technical communities.
The article argues that the proliferation of AI-generated content (slop) is causing a provenance crisis where the origin and reliability of information are undermined, illustrated by examples of misdirected automated outreach and fake engagement.