The "output rating" positive feedback loop is very concerning to me. That is a huge, growing advantage for the incumbants. Government can't necessarily force the publication of customer queries/responses to public datasets.
But there are ways for open source to compete, like Artbot's opt in contribution and rating system for LAION:
But it only works if community projects and small companies pile into common transparent datasets (rather than trying to create tiny silos and reinvent the wheel like most do).
And I think many "anti AI" communities (like writing and art websites) should reconsider the implications of shutting off access to open source training
But there are ways for open source to compete, like Artbot's opt in contribution and rating system for LAION:
https://tinybots.net/artbot/rate
But it only works if community projects and small companies pile into common transparent datasets (rather than trying to create tiny silos and reinvent the wheel like most do).
And I think many "anti AI" communities (like writing and art websites) should reconsider the implications of shutting off access to open source training