Hold up though, when some of these companies go bankrupt and their assets get sold off for cheap....
One of the main objections I have to "large model" AI is that it is a form of piracy that fails to serve as an accurate form of archival. https://vis.social/@bleeptrack/117560078386333325
Every AI scraper develops an archive of copyrighted works that would put any library on earth to shame. Then they only let you access it it horrifically garbled, with semantic noise added, and through a straw
@mhoye Not totally clear to me how diligently they keep the material between training runs. Like, it might make the most sense to do so, but they do things that don't make sense