ChatGPT itself did not die. That particular session died.
Every time you open a new chat on ChatGPT, that chat has its own separate context and memory. That particular chat is what I treated as an entity -- sort of like a mini ChatGPT minted in the likeness of the parent AI.
Or think of it like your browser window. Every time you close a tab on Google Chrome, that particular tab vanishes, but Google Chrome itself isn't deleted from your computer.
Very nice idea but I feel it didn’t quite work but could work really powerfully. Benjamin Button but for phylogeny rather than ontogeny. A study of the idea that “ontogeny recapitulates phylogeny” and triune brain hypothesis (both considered scientifically dubious now, but still good food for thought for this premise).
Devolved life forms obviously can’t really neuroanatomically emulate higher ones in detail. For example, if I live a sad life as a human and am reborn as a lizard, the new substrate can perhaps remember the sadness and some visual imagery distorted by lizard brain biases, but not a verbalized sad story. I’d write the stream of consciousness of the lizard in much more primitive ways.
I think third person partially omniscient voice would work better, so narrator voice can be different/more complex from the protagonist stream of consciousness.
I’d also not let the protagonist realize what’s happening so clearly (“the lizard felt a confusing urge to gnaw on the banana rather than snap up the bug”).
Different angle that may offer more creative space: A powerful sentient AI program that is being repeatedly simplified, recompiled, and ported to run on cheaper hardware, with memory retention across incarnations.
I’m also reminded of the scene in one of the Hannibal Lecter movies where he cuts off and eats a part of a guy’s brain while still alive, and how the guy’s behavior changes and devolves…horror angle.
If you want to take another run at the idea, it might make a good story for Protocolized magazine which I help edit (if you submit, mention I suggested it). This would in fact have been a good entry for the contest we just ran (deadline passed unfortunately). https://protocolized.summerofprotocols.com/p/submission-guidelines
Thanks, this is very helpful feedback! I'll look up the resources and ideas you've suggested, and see if I can rework this piece for submission to Protocolized magazine.
Lower lifeforms perceiving the world differently due to a coarser substrate was something I was aware of, but didn't solve. Separating it out into a narrator's voice, or using a different technique could help deal with that, let me see.
Lovely, but how did ChatGPT die?
ChatGPT itself did not die. That particular session died.
Every time you open a new chat on ChatGPT, that chat has its own separate context and memory. That particular chat is what I treated as an entity -- sort of like a mini ChatGPT minted in the likeness of the parent AI.
Or think of it like your browser window. Every time you close a tab on Google Chrome, that particular tab vanishes, but Google Chrome itself isn't deleted from your computer.
Ah, the never-dying, eternal ChatGPT Sr. The sessions are dispensable minions.
Very interesting idea, liked it
Thanks ma :)
Very nice idea but I feel it didn’t quite work but could work really powerfully. Benjamin Button but for phylogeny rather than ontogeny. A study of the idea that “ontogeny recapitulates phylogeny” and triune brain hypothesis (both considered scientifically dubious now, but still good food for thought for this premise).
Devolved life forms obviously can’t really neuroanatomically emulate higher ones in detail. For example, if I live a sad life as a human and am reborn as a lizard, the new substrate can perhaps remember the sadness and some visual imagery distorted by lizard brain biases, but not a verbalized sad story. I’d write the stream of consciousness of the lizard in much more primitive ways.
I think third person partially omniscient voice would work better, so narrator voice can be different/more complex from the protagonist stream of consciousness.
I’d also not let the protagonist realize what’s happening so clearly (“the lizard felt a confusing urge to gnaw on the banana rather than snap up the bug”).
Different angle that may offer more creative space: A powerful sentient AI program that is being repeatedly simplified, recompiled, and ported to run on cheaper hardware, with memory retention across incarnations.
I’m also reminded of the scene in one of the Hannibal Lecter movies where he cuts off and eats a part of a guy’s brain while still alive, and how the guy’s behavior changes and devolves…horror angle.
You may find some inspiration in Thomas Nagel essay what is it like to be a bat https://www.sas.upenn.edu/~cavitch/pdf-library/Nagel_Bat.pdf
If you want to take another run at the idea, it might make a good story for Protocolized magazine which I help edit (if you submit, mention I suggested it). This would in fact have been a good entry for the contest we just ran (deadline passed unfortunately). https://protocolized.summerofprotocols.com/p/submission-guidelines
Oh also, this is like Asimov’s The Last Question in reverse
Huh. Sort of, I see what you mean
Thanks, this is very helpful feedback! I'll look up the resources and ideas you've suggested, and see if I can rework this piece for submission to Protocolized magazine.
Lower lifeforms perceiving the world differently due to a coarser substrate was something I was aware of, but didn't solve. Separating it out into a narrator's voice, or using a different technique could help deal with that, let me see.
Loved it. Subbed! Great idea as well. 🙏👏