Comment Re:The American Dream (Score 2) 18
Outright lying about your product though... that's bad enough and opens you up to prosecution. But lying about the numbers? That is a big no-no... They will come after you for that, with a vengeance.
If I read copyrighted material and then write my own story using information in what I read is that infringement?
It depends. If you wrote the same story, but with just the names of characters and places changed, then yes. In general, it depends on how similar your story is to the original.
Do clif notes have to pay for the rights to publish notes on other authors books?
No. Their notes were written by them and so are copyrighted by them. If they give excerpts, that falls under fair use. They can't quote the entire book, however, even if they give notes on every paragraph.
Isn't this is what ChatGPT does?
There are two parts to this: whether the copy of what CharGPT was trained on was a legally obtained copy in the first place and whether ChatGPT can be induced to regurgitate verbatim copies. The NYT demonstrated that, with the right prompting, ChatGPT can be induced to regurgitate copies of NYT articles.
Once I learn about something I am able to tell people what I have learned. Am I breaking the law if I tell my grandkids about a story I read.
It depends how you tell them. If you give, for example, highlights, then no. If you write down a copy and give them the copy, then yes. Also note that, assuming you either paid for your copy of the book or you borrowed it from a library, then the copy you used was a legally authorized copy to begin with.
I just don't see a lot of difference between asking an AI a question and having it tell me what it has learned than asking a human and having them tell me what they learned assuming they were trained on the same materials.
In many cases, ChatGPT used unauthorized copies to begin with. At that point, they're already guilty of copyright infringement. If the book the human read was an unauthorized copy, then that human is guilty of copyright infringement. Whether they tell you anything is irrelevant: they're already guilty.
Assuming both legally obtained the info they were trained on.
Bingo.
ChatGPT doesn't copy most works
It copies every work verbatim. It has to have a copy in the first place to train from. If it made its copy from an unauthorized copy to begin with, then it's already guilty of copyright infringement. At this point, the training is irrelevant.
For example, if its web crawler came upon an unauthorized PDF of a book that somebody uploaded to some web site, then the uploader and every downloader has committed copyright infringement. What they do with the copies is irrelevant: they're already guilty.
I don't believe the government has the right to ban me from harming myself in their eyes.
If you signed a contract stating that you agreed either to forgo any treatments for health issues arising from your choice to smoke or agreed to pay for all such treatments out of your own pocket, then fine. But, and especially if you live in a country with socialized medicine, then everyone else is paying for your treatment. So to minimize everyone's expenses for your completely avoidable health issues, the government is doing what it's doing. And they haven't stopped you from harming yourself. They just put images and words on the package.
Using TSO is like kicking a dead whale down the beach. -- S.C. Johnson