• Grimy@lemmy.world
        link
        fedilink
        arrow-up
        3
        arrow-down
        6
        ·
        21 hours ago

        I’m sure blind people are happy to have the models that are built with this data, and since both the image and the description are public facing, anyone can use them including open source.

        • achille225
          link
          fedilink
          arrow-up
          4
          ·
          8 hours ago

          Except you need to pay Reddit to use their data for training

          • Grimy@lemmy.world
            link
            fedilink
            arrow-up
            1
            ·
            1 hour ago

            No. That is what data brokers and big AI companies are pushing for but currently it’s considered fair use.

            Anything public facing can be used for ml and it’s been like that for quite a while. It might change based on all the ongoing lawsuit but I doubt it will, it would be economic suicide and China doesn’t care if it’s “theft”.

            It’s better for us, the consumer in any case, since having to pay for data would kill the open source scene and give openai and the other 3 companies a defecto monopoly.