I’m still of the opinion that we should be allowed to train on any data a human can read online.
…Limited to a human’s average rate of consumption, right?