AI Watchdog Datasets

AI Watchdog: HowTo100M

Editor’s note: This search tool is part of The Atlantic’s investigation into how YouTube videos are taken to train AI tools. This work is part of AI Watchdog, The Atlantic’s ongoing investigation into the generative-AI industry.

A collection of 1,238,911 YouTube videos split into more than 100 million video clips. The clips show 23,611 “tasks” involved in cooking, handcrafting, personal care, gardening, and other activities. Task names and descriptions were based on WikiHow’s task taxonomy. The data set is hosted on a website of the European research institutions that created it.