overgeeked
Open-World Sandbox
No on all counts. There are too many living humans writing amazing things to want to pseudo-resurrect a dead game designer to produce more works. Besides, it's recursive. Training an AI on his works would only rehash and remix what exists, not create anything remotely new.So, would you buy such a work? Would you buy it if the proceeds went to his estate? Would you buy it if the proceeds went to just the people publishing it?
Literally none of them. Ever. Again, too many living humans writing amazing things to want to bother training an AI to produce fake works for consumption. The stack of existing literature is already so overwhelming that it would take several human lifetimes 100% dedicated to nothing but reading to get through even a tiny fraction of all available literature.And what other works along these lines would you buy, once the technology was good enough that the quality was on par with a human author's below-average day? Would you buy The Winds of Winter by an AI-generated George R. R. Martin? Would you buy new Lord of the Rings works by an AI-generated Tolkien?
Exactly.no, get an actual writer to go through the notes and fill in the gaps / add to it, and an artist to create maps, etc.
same, no, there is enough ‘real’ literature out there that I have no interest in reading the mechanized drivel of an AI, even if it nails the writing style.
I have read some fanfiction. In most cases it's below average quality-wise compared to the original. In a very few, incredibly rare cases, the fanfiction is better than the original. Just like how every so often a cover song can be better than the original.I do not read fan fiction either however, so there probably are some people out there that would
Training AI on copyrighted works very clearly requires making a copy of the original work without permission, thus violating copyright. Many of the programmers involved got hold of the texts they used to train their AIs by pirating those works from torrent sites...which is another copyright violation. Even if that were not the case, generating new works using the original author's IP without permission (i.e. unauthorized derivative works) is also very clearly a violation of copyright. There's no way around it. Generative AI violates copyright at every step of the process.I also do not like this bit
“Training AI on copyrighted works isn’t actually illegal. If the real Martin had wanted to block access to the fake one — a replica trained on his own thinking, using his own words, to produce all-new answers — it’s not clear he could have done anything about it.”
It absolutely should be illegal