Dan Romero pfp
Dan Romero
@dwr.eth
Assume an LLM trained on your favorite writer / fiction series was indistinguishable from the original. Would you spend time reading infinite content in that literary universe? i.e. The Foundation or Dune but the ability to zip backward or forward in the timeline
25 replies
0 recast
0 reaction
Pedro pfp
Pedro
@pedrosilva.eth
maybe have a chat with the characters?
0 reply
0 recast
0 reaction
John Halloran pfp
John Halloran
@jh
Related q: which writer/series’ corpus creates the most engaging LLM
0 reply
0 recast
0 reaction
John Halloran pfp
John Halloran
@jh
Imagine it could be fun to interact 1 on 1 with multiple characters; get more perspective on plot points
0 reply
0 recast
0 reaction
gidwell 🎩 pfp
gidwell 🎩
@gidwell
I guess this is the only way we finish the Song of Ice and Fire books cc George
0 reply
0 recast
0 reaction
DBKW🦉🎩 pfp
DBKW🦉🎩
@drinkbeerkillwar.eth
@perl #hipporeader
0 reply
0 recast
0 reaction
jcdenton.cast pfp
jcdenton.cast
@jc
@beb
1 reply
0 recast
0 reaction
moreReese pfp
moreReese
@morereese
No bc LLM would train on the written content (vocabulary, sentence structure, etc) so I think the imaginative appeal would asymptote. Especially considering how much mimicry there is in fiction writing. Would be super interesting tho if the LLM also trained on the author’s exhaustive biography or diary
0 reply
0 recast
0 reaction
zom pfp
zom
@zom
Yes Would need some form of ‘canonizing’ system to sift the best and most cohesive pieces into an official weave This system could also be AI
0 reply
0 recast
0 reaction
Sam Iglesias pfp
Sam Iglesias
@sam
that’s a very loaded premise, but assuming the system realistically emulated the author drawing from and responding to current politics, events, zeitgeist, YES
0 reply
0 recast
0 reaction
_slow_crypto 🌅⏳🪁 pfp
_slow_crypto 🌅⏳🪁
@slowcrypto
Some, sure, but certainly not infinite (or even a large amount). That’d become super boring. I want new universes to be excited about too. And more so, I want a world that inspires and incentivises writers to invent them!
0 reply
0 recast
0 reaction
Alex pfp
Alex
@alexyao
I want to read the backstories of various universes of my fav sci-fi novels and generated storylines of different characters
1 reply
0 recast
0 reaction
Nicholas Charriere pfp
Nicholas Charriere
@pushix
anything to get volume 3 of the Kingkiller Chronicles.
1 reply
0 recast
0 reaction
Christopher McElveen pfp
Christopher McElveen
@cmac
I explored something similar this weekend. Basically kids ask questions and are curious. They like to be read to. Use chatGPT to write an original short story and enable the parent or kid to either extend the story back or forward in time or expand on a particular character or sub plot. V interesting potential there!
0 reply
0 recast
0 reaction
czar pfp
czar
@czar
Beautiful question. I think a major part of what makes a content great is also the social aspect of it. As long as the content: a) has consistent storyline irrespective of the volume of content b) has at least some high quality b) and also social consensus on what’s high quality Why not? Throw me some LLM LOT
0 reply
0 recast
0 reaction
mk pfp
mk
@mk
Maybe. I can imagine doing so, but it might also lose the magic for me.
0 reply
0 recast
0 reaction
manansh ❄️ pfp
manansh ❄️
@manansh
More Enders game please
0 reply
0 recast
0 reaction
Ertan Dogrultan pfp
Ertan Dogrultan
@ertan
On a similar note, soon it might be possible to generate continuous steam of Interdimensional cable content from Rick and Morty.
0 reply
0 recast
0 reaction
July pfp
July
@july
Nope. I wouldn’t
2 replies
0 recast
0 reaction
Integrated Kyle e/acc pfp
Integrated Kyle e/acc
@kylesamani
i wonder if statistical models in LLM can pick up the nuance of what creates emotional engagement to characters over a long arc of a story. my sense is that right now, they can't
1 reply
0 recast
0 reaction