#calling out AI isn't “starting drama” it's about upholding fandom to a certain standard

[elf-trash]

reposting this bc the OP blocked me (and is blocking anyone else who disagrees which means blocked people can't reblog) and i want to say this loud and with my whole chest!!!!!

another Dragon Age fic was recently outed as being AI, and this is what the writer had to say for themselves about it:

so actually, Grammarly uses generative AI and is just as bad as ChatGPT. it also objectively makes your writing worse, it sucks the voice out of your prose and turns it into corporate sounding homogenized paste. it's also unethical for all the same reasons any generative AI is unethical. get a writing group and have a real human beta read for you if you don't trust yourself to check your own grammar etc. but honestly something unpolished and written entirely by your human brain and human imagination will ALWAYS be better than AI slop.

also, the part about published authors doing this is patently untrue. i know this is a huge problem in the self-publishing space, but most publishers now are including clauses in their contracts that expressly forbid the use of AI in ANY part of the creative process. this includes using ChatGPT to generate or clean up outlines or Grammarly to spellcheck and revise. so if you're trying to publish, don't fucking do this or you could literally be asked to return an advance if you get caught.

i've posted about this in the past, but AI detectors are actually shocking accurate these days. i've tested them extensively recently and they can consistently and correctly flag individual sentences written by ChatGPT in an otherwise original passage. and they almost never flag false positives. so the argument that AI detectors can't be trusted is just flat out wrong. are they correct 100% of the time? no. but can they indicate with a high degree of accuracy if AI was used in some capacity? absolutely, especially if there is additional evidence.

and for all the people hand wringing about AI detectors flagging false positives, let me just say this: if something is not AI written it is very easy to prove. you can't write anything of any considerable length without leaving a massive paper trail of notes and drafts. almost all writing software tracks changes and makes it very easy to prove you wrote something yourself. being falsely being accused of AI isn't actually a real problem and is only being made to seem as such by people who are trying to get away with and justify using AI or who are worried about getting caught.

i think a lot of people are just lured by a seemingly easy shortcut, and to their untrained eye, what the AI is spitting out feels "better" to them than their own writing. but i promise you it's not. trust your own brain and put in the work to improve at your craft rather than outsourcing the gift of your imagination to a robot that steals from other people's work.

i will continue to die on this hill!!!!!

[elf-trash]

actually Grammarly uses generative AI and is just as bad as ChatGPT. it also objectively makes your writing worse, it sucks the voice out of your prose and turns it into corporate sounding homogenized paste. it's also unethical for all the same reasons any generative AI is unethical. get a writing group and have a real human beta read for you if you don't trust yourself to check your own grammar etc. but honestly something unpolished and written entirely by your human brain and human imagination will ALWAYS be better than AI slop.

also, the part about published authors doing this is patently untrue. i know this is a huge problem in the self-publishing space, but most publishers now are including clauses in their contracts that expressly forbid the use of AI in ANY part of the creative process. this includes using ChatGPT to generate or clean up outlines or Grammarly to spellcheck and revise. so if you're trying to publish, don't fucking do this or you could literally be asked to return an advance if you get caught.

i've posted about this in the past, but AI detectors are actually shocking accurate these days. i've tested them extensively recently and they can consistently and correctly flag individual sentences written by ChatGPT in an otherwise original passage. and they almost never flag false positives. so the argument that AI detectors can't be trusted is just flat out wrong. are they correct 100% of the time? no. but can they indicate with a high degree of accuracy if AI was used in some capacity? absolutely, especially if there is additional evidence.

and for all the people hand wringing about AI detectors flagging false positives, let me just say this: if something is not AI written it is very easy to prove. you can't write anything of any considerable length without leaving a massive paper trail of notes and drafts. almost all writing software tracks changes and makes it very easy to prove you wrote something yourself. being falsely being accused of AI isn't actually a real problem and is only being made to seem as such by people who are trying to get away with and justify using AI or who are worried about getting caught.

i think a lot of people are just lured by a seemingly easy shortcut, and to their untrained eye, what the AI is spitting out feels "better" to them than their own writing. but i promise you it's not. trust your own brain and put in the work to improve at your craft rather than outsourcing the gift of your imagination to a robot that steals from other people's work.

i will continue to die on this hill!!!!!