It’s official: The robots are taking up. Taking on the web, that’s.
Conspiracy theorists have lengthy mentioned the “useless web” concept, which causes that on-line areas, as soon as totally populated by and full of content material created by people, have slowly turn into dominated by bots posing as individuals. The extra excessive conspiracists allege that this transformation is deliberate, with governments and companies utilizing the bots to govern public notion.
With the rise of AI since ChatGPT’s debut in 2022, the useless web concept—or no less than some model of it—has sounded an increasing number of believable. Now, based on a latest examine, it’s nearer to coming true.
The study, a collaboration amongst researchers at Stanford College, Imperial School London, and the Web Archive, sought to search out how a lot textual content on the web is AI-generated. To take action, they used the Web Archive’s Wayback Machine to check net pages printed from 2022 to 2025, utilizing a number of distinguished AI-detection strategies.
Their outcomes present that as of Might 2025, greater than a 3rd (35.3%) of all new web sites had been AI-generated or AI-assisted. That features 17.6% of all newly printed web sites being totally generated by AI.
The examine’s findings corroborate different knowledge, together with Cloudflare’s report that just about a 3rd of all web site visitors over the previous yr comes got here from bots, and Imperva’s claim that 2024 noticed automated site visitors surpass human site visitors for the primary time.
“I discover the sheer pace of the AI takeover of the net fairly staggering,” Jonáš Doležal, one of many examine’s researchers, informed 404 Media. “After a long time of people shaping it, a good portion of the web has turn into outlined by AI in simply three years. We’re witnessing, for my part, a serious transformation of the digital panorama in a fraction of the time it took to construct within the first place.”
Defying AI stereotypes
Whereas AI-generated web sites have gotten an increasing number of frequent, they will not be as dangerous to the web as critics concern. The examine additionally examined AI-generated content material’s impact on the web through six hypotheses centered on frequent critiques of AI-generated textual content, solely two of which had been confirmed by the analysis.
The examine confirmed that AI is contributing to semantic contraction, or the discount of numerous viewpoints on-line, and to a positivity shift, via which on-line writing is general turning into extra sanitized and artificially cheerful.
However to this point, AI is seemingly avoiding different suspected detrimental results. The examine didn’t see proof of will increase in rambling textual content with little to no substance, a single generic writing fashion, an absence of cited sources, or—maybe most shockingly—the unfold of misinformation from AI-generated content material on-line.
The examine’s authors informed 404 Media they’re working to show their analysis right into a steady instrument in order that web customers can keep updated on simply how useless that “useless web” actually is.
“As AI-generated content material spreads, the problem is discovering a job for these fashions that doesn’t simply end in a sanitized, repetitive net,” Doležal mentioned. “Somewhat than forcing fashions to be completely compliant and agreeable, permitting them to have a extra distinct persona or ‘friction’ may assist them act as a artistic companion slightly than a substitute for human voice.”
