A Third of New Web Pages Are Written With AI: Filtering Synthetic Content Out of Scraped Datasets

Dark teal cover with the post title and a diagram of five scraped records, each with an AI-score bar, a dashed threshold line, and the two records above it flagged.

Pew estimates 35% of new dated web pages show AI authorship. Here is how to filter AI-generated content from scraped datasets without fooling yourself.

Tag: datasets

A Third of New Web Pages Are Written With AI: Filtering Synthetic Content Out of Scraped Datasets

Request this dataset