8
That AI art panel at SXSW last month completely flipped my view on training data
Honestly, I went in thinking all these lawsuits over scraping were overblown. But the speaker from a mid-size studio showed us a breakdown of where their model got 40% of its source images from. It was all uncredited work from DeviantArt and private portfolios. Has anyone else actually dug into where their favorite AI tool pulls its training set from?
3 comments
Log in to join the discussion
Log In3 Comments
jordan_webb491mo ago
Clicked through a few of those links after the panel. Turns out one of my favorite "retro filter" apps basically just rebuilt someone's whole photography portfolio without asking.
1
the_morgan1mo ago
Used to wave off the arguments about data scraping as no big deal. Seeing actual portfolios getting repackaged like that changed my mind quick.
3
brookekelly1mo ago
Wait, someone actually rebuilt their whole portfolio from it... that's messed up.
4