Picture of book cover SHY GIRL by Mia Ballard Text: Test #2 With the AI scandal erupting regarding Mia Ballard’s book, SHY GIRL, I thought I would conduct an experiment. How many of my writings would be flagged as AI. To keep it above board, I will only use my fanfic stories I wrote between 2003 -2016, before AI generated writing came to be. Scroll to see the results
This second fanfic I am testing was also written in 2004, like test #1 sample. This one is a parody of bodice ripper romance novels and uses bad metaphors. So we’ll see if the purposeful use of bad metaphors for humor flags this as AI. Screenshot from AO3 showing it uploaded from FictionAlley in 2015. Screenshot from Ashwinder showing it uploaded in 2004.
Screenshot of results from Pangram, Quillbot, JustDone, CopyLeaks, ZeroGPT, GPTZero, and Grammarly.
RESULTS: Pangram: 100% human CopyLeaks: 0% AI ZeroGPT: 32.8% AI GPTZero: 100% human, 0% AI Quillbot: 94% human, 6% AI refined Grammarly: 0% AI JustDone: 88% AI content
CONCLUSION: Test #2
Compared to the test sample #1, JustDone, Quillbot and ZeroGPT, scored the use of AI even higher, with test sample #2. ZeroGPT made a point of highlighting the bad metaphors and purple prose as proof it was written by AI.
#AI #ShyGirl #AIscandal #ShyGirlScandal #WritingCommunity