GOG Wiki

Topic

AI training data licensing

15 mentions in 6 episodes

Also known as AI training data

The hosts discuss a European Parliament-commissioned report proposing statutory licensing and centralized royalties for AI training data.

Mentions

726: Sycophantic and Delusional

  • 16:59

    treating AI training data like the music industry handled Napster is a doomed strategy

  • 17:15

    The proposed fix is statutory licensing

  • 17:52

    they're saying no opt outs allowed

660: The Bendy Piece

  • 11:19

    Just because you put it on the internet means we can train on it.

  • 11:35

    NVIDIA has been using videos from YouTube, Netflix, and other sources to train a new AI model

  • 14:19

    opt everybody in to training Grok with everything that they do on the X platform

657: Dick Detector

  • 16:50

    using YouTube transcripts without permission to train its AI models

  • 17:45

    data siphoned from creators without their consent or compensation

652: Casual Pirates

  • 4:10

    They're saying that by 2026, they're going to run out of shit.

  • 5:03

    it then gets trained on

643: Mechanical Goslings

  • 53:32

    AI companies are trying to make larger and larger models.

  • 54:13

    They're thinking about scraping public YouTube videos and getting the transcripts.

  • 54:45

    they're debating actually putting in synthetic data, which is generated by the AIs.

599: Osaka

  • 12:04

    start charging these companies that want to train their AI models

  • 12:19

    these other companies who take your content, post it, and then resell it