from Hacker News

OpenAI partners with Vox Media, Anthropic brings ex-OpenAI safety lead

by arihantparsoya on 5/29/24, 4:49 PM with 9 comments

  • by tombert on 5/29/24, 5:26 PM

    I am pretty much every stereotype of a lefty American, but even I have some hesitation of using Vox media to train AIs. Vox is pretty non-apologetically partisan, which isn’t necessarily a problem but I don’t know is a good idea for training AI.

    I do actually really like Vox, but I dunno, this is giving me a bad vibe.

  • by Imnimo on 5/29/24, 5:31 PM

    I'm still unclear whether these partnerships are "we're going to make a special segment of our training corpus that is text from Vox Media because we think it's extra valuable" or "we almost certainly scraped a bunch of Vox data when we built our training corpus years ago, and now we're paying them to not sue us".

  • by waterhouse on 5/29/24, 5:42 PM

    > OpenAI partners with Vox Media

    This strikes me as very weird, coming after some highly critical reporting by Vox about OpenAI: https://news.ycombinator.com/item?id=40447431

  • by ChrisArchitect on 5/29/24, 9:52 PM