• nucleative@lemmy.world
    link
    fedilink
    English
    arrow-up
    7
    arrow-down
    1
    ·
    16 hours ago

    I don’t automatically have a negative opinion about this, I would need more information before that. Did the terms of service allow for this?

    It’s a fascinating case study on crowdsourcing data that is useful to this navigation technology, and reminds me of the first captchas that helped train OCR engines.

    • frongt@lemmy.zip
      link
      fedilink
      English
      arrow-up
      4
      arrow-down
      1
      ·
      15 hours ago

      Yeah this is like 90% of all of Google’s business model. Captchas to train OCR or tag crosswalks, vehicles, etc. for other computer vision. GOOG-411 to train voice recognition for the Google assistant.

      • paraphrand@lemmy.world
        link
        fedilink
        English
        arrow-up
        3
        ·
        edit-2
        15 hours ago

        Google voice only makes sense as a data play too, for voicemail data, etc.

        YouTube is a massive dataset. Part of their recent crackdowns on downloading tools and 3rd party apps is around keeping that data for themselves.