Honestly maybe (and just maybe) waiving copyrights on all the written content that ever existed to create training data would be a good thing to do, laws are made up so we can decide it’s a good trade off as a society. But:

1. I feel like this should be discussed globally, there should be a public debate, a vote, and guardrails

2. It should not be in the hands of private companies, it should either be done by the government and made available to the public ; or if it’s done by private companies they should be mandated to give the training data to the government so it’s available to the public.

My point is we can decide to say it’s ok because LLMs are too important strategically. But if we do so it should benefit the public, not 5 mega corporations, training data should be considered as public infrastructure, like roads, rails, or the electricity grid. Societies are failing and this is just one more nail in the coffin.