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@jlai.lu 5 points 3 weeks ago (1 children)

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

[–] Grimy@lemmy.world 1 points 3 weeks 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.