←back to thread

247 points inesranzo | 2 comments | | HN request time: 0.015s | source
Show context
shadowgovt ◴[] No.46232889[source]
For everyone concerned about the AI systems being trained on copyrighted material: this was always the end-game of that argument. Once the technology was proven out to be useful, someone with a huge IP portfolio was going to slam that portfolio directly into the training data to get their own copyright-unencumbered AI.
replies(1): >>46233419 #
dfedbeef ◴[] No.46233419[source]
Copyright unencumbered... For their own characters?? Why would they need clearance to generate things trained on their data.
replies(1): >>46235154 #
1. shadowgovt ◴[] No.46235154[source]
Because the tool can generate output faster than human employees alone can (and they control the "legitimate" gate by which other people can use those characters in the tool).

The consequence being that for everyone complaining that AI is disrupting artists right now: these will, in hindsight, be the halcyon years. Even if we assume the copyright arguments hold water in court and AIs trained on other people's copyrighted material are ruled poison-fruit machines, the end result isn't the end of synthesizing-AIs... It's synthesizing-AIs only being owned by people with a big enough data portfolio to train them. Techno-anarchy replaced with techno-corporatocracy, and the smaller-volume artists still lose on being unable to out-produce their competition in an art market.

replies(1): >>46243241 #
2. dfedbeef ◴[] No.46243241[source]
Art isn't a volume business though.