We Thought AI Would Fill the Internet With Lies.

Three studies by researchers from Stanford, Imperial College London, and the Internet Archive found that AI-generated content now accounts for 35% of new websites, and while it hasn't made the internet less accurate, it is reducing diversity of information and homogenizing online content.

Member-only story We Thought AI Would Fill the Internet With Lies. Scientists Just Measured What’s Actually Disappearing AI Isn’t Making the Internet Less Accurate. It’s Making It Less Diverse.Three Studies Just Proved It. Three years after ChatGPT, 35% of new websites are AI-generated. Researchers from Stanford, Imperial College London, and the Internet Archive spent 33 months measuring what AI content does to the web. The results contradict almost everything the public believes — and reveal a threat nobody was watching for. Try this experiment. Open a search engine and ask a question you genuinely care about — something specific, something that matters to your work or your thinking. Then open five of the results. Read the first three paragraphs of each. Chances are you will notice something difficult to name but impossible to miss. The answers are not wrong, exactly. They are comprehensive, well-structured, and optimistic. They cover the main points. They reach broadly similar conclusions in broadly similar language. They feel, in some way that…