[flagged]
I think the point of bans is to keep TOS-violating content off of platforms, not to rehabilitate people.
I’ve got no evidence, but I would like to believe that Twitch has experimented with ban lengths and found 14 days to be enough to reduce TOS-violating content but not so much that it reduces total content.
If you stream a video and showcase a comedian making ToS-violating jokes and content, twitch will ban you regardless of whether you're a person or a robot or a script or a fish in a tank.
The only difference is if you're a major company or rights holder that Twitch wants to have a good relationship with. Or if you're a person with connections to the twitch mods and admins, they have a tendency to play favorites. But being AI generated content or not really isn't the issue here.
So.... Machine Learning is a thing.
> That's the whole point behind bans, is it not?
No, it may be one point behind imposing bans but it is not the whole (only) point. People can also learn from watching other's make mistakes. Bans communicate to people other than the person getting banned that something is unacceptable.
In this case it seems to communicate that "but it is not a person it's a bot" will not be a srong enough reason to break their rules and escape a ban.
A ban is just a message to the creator that it should only come back once it has figured out how to properly label its dataset so that the training complies with the ToS.
Since when?
I thought it was to allow rightthink and kick out wrongthink so we may curate areas in which the latter has been filtered out, a filter bubble if you will...