Much of the newer web appears to be machine-written. A Pew Research report released August 20 found that more than a third of English-language pages published since ChatGPT’s debut carry signs of AI authorship or heavy AI editing.
To reach that figure, Pew drew nearly half a million pages from the Common Crawl archive spanning roughly five years, starting before ChatGPT’s November 2022 launch. Detection technology from Open Pangram was applied to a random sample of 10,000 pages to estimate how many were generated or substantially edited by AI.
The study lines up with earlier research and with Cloudflare’s recent report that bot traffic has overtaken human traffic. Pew’s data focuses on what is being browsed rather than who is browsing, and the picture it paints is of machines reading pages written by other machines.
The report is a reminder that the web’s information layer is becoming synthetic at a remarkable pace.
