I’m not asking if they gave up copyright on their photos. They did. I’m asking, for example, if it’s ethically appropriate for Microsoft to publish annotated pu

I’m not asking if they gave up copyright on their photos. They did. I’m asking, for example, if it’s ethically appropriate for Microsoft to publish annotated public domain photos without requiring a human ethical review for each use of their dataset. If I wanted to perform a sociological study on that dataset, I’d have to get a review board’s approval. Why is performing a statistical study (literally, machine learning) somehow exempt from that ethical concern?