On: NY Times copyright suit wants OpenAI to delete all GPT instances

[fuzzfactor]

What if you were one of the people who read the Times from cover-to-cover every day and seriously tries to remember as much as possible because you consider it a trustworthy reference source?

And if you were called upon to solve a problem based on knowledge you consider trustworthy, what would you come up with?

What if you were even specifically directed to utilize only findings gleaned from the Times exclusively?

And what if that was your only lifetime source of information whatsoever for some reason?

[jjeaff → fuzzfactor]

That would of course be fine.

But then imagine that because human memory is not able to keep all that information straight, you made copies of all those newspapers.

And then you started charging people for your knowledge.

And then imagine that as part of your knowledge service, you would copy snippets from the times word for word and give that to your clients without citation and pass it off as your own.

[fuzzfactor → jjeaff]

Yup, that's the other side of the dodecahedron.

As I understand it, it's the copying that can lead to infringement.

Then again if you have acquired a legitimate copy, you should certainly be able to retain it and use it for reference.

But for training a model on someone else's data I wouldn't even want a copy.

Just skim the data and retain my own thoughts.