One more thing: if you want to use public data, your AI needs to be open source (not just the software around it, the actual models that do the AI stuff needs to be available for anyone to run on their own system) and all the works generated with it public domain. The public owns your AI at that point. Personally, if you don’t want to pay me, then let me have a stake in the AI my data helped create.
Whenever I hear “but how can the police be racist if there are so many black officers?!” I think of this line from NWA: