zlacker

From their perspective they're training a giant mechanical brain. A human brain doesn't need any special license agreement to read and learn from a publicly available book or web page, why should a silicon one? They probably didn't even consider the possibility that people'd claim that merely having an LLM read copyrighted data was a copyright violation.

>>logicc+(OP)
I was thinking about this argument too: is it a "license violation" to gift a young adult a NYT subscription to help them learn to read? Or someone learning English as second language? That seems to be a strong argument.

But it falls apart because kids aren't business units trained to maximize shareholder returns (maybe in the farming age they were). OpenAI isn't open, it's making revolutionary tools that are absolutely going to be monetized by the highest bidder. A quick way to test this is NYT offers to drop their case if "open" AI "open"-ly releases all its code and training data, they're just learning right? what's the harm?