I like the first section; I think you wrote the first section. It gets increasingly cringy ongoing but I wonder if there is also a factor of the diff models only writing a piece?
Yeah, the “handover” effect is definitely a factor here. I'm sure one of the better LLMs would've done a better job on its own than in this combo-writing experiment.
I also quite like the first section, but I won't reveal whether I wrote it just yet. Let the people guess!
Also, remember that my goal is to try to blend in, so liking a section might not necessarily mean it's mine.
I vote section 8 human hand, the squeeze ball is confusing if not a human (why does a robot have a stress toy), I think the woman reference is too illogical for a machine
Yeah, mimicking an accent is right up ChatGPT's "parody" alley. I mentioned that wordplay, parody, and analogy are the easiest types of humor for AI here: https://www.whytryai.com/i/147880162/ai-excels-at
I'm pretty sure the first part was ChatGPT, based solely on the name "Elian". It's used that name in a sci-fi story for me recently, and I've noticed it tends to often use the same or very similar names across different stories.
I think 2 is AI as well. "Humans are messy" is the kind of thing AI's like to say. I once had two GPTs pretend they were a human administering a Turing test on the other, and it was full of them saying back and forth how "messy" and "real" and "raw" each other's responses were.
3 has to be AI. "Now, its grief isn’t sadness. It’s silence." Unless you included that as an intentional curve ball...
I'm much less confident for the rest, but if I had to guess, I'd say number 6 is you. I'm not sure exactly why. I think the number in 7 is probably a sign it's AI, but I'm not sure. 6 also seems to return to about the right section length, whereas the others had been getting gradually longer.
I enjoyed the whole story. I'm OK with a little slop, just like I'm OK with a little fast food.
I like using AI to do "Choose Your Own Adventure" stories, like I used to occasionally see as a kid. The kind where at the end of each page it would ask how you want to respond, and tell you what page to turn to for it. They were "slop" too, but it's fun and I don't care.
Nice! Concrete guesses backed up by experience - love it!
My very first attempt at making a Custom GPT was "Quest Weaver," which was a CYOA-style story maker. I also wanted to add RPG elements where people could create a character with attributes, and the story would be affected by those (and their decisions). Then DALL-E 3 would generate images for the scenes and locations. It didn't quite work at the time, as I described here: https://www.whytryai.com/p/build-gpts
But that was ages ago, during the early days of GPTs, so maybe things have improved.
I originally planned to only briefly reveal who wrote which parts of the story, but now that there are so many guesses and opinions, I'll probably expand it into a full post with some observations about AI fiction.
Without hopefully spoiling too much, I'd like to point out that there's no conflict in #2's passage and the original premise. I see at least two possible interpretations here, under both of which the premise still holds:
1. The "man" in the suit isn't human. The text never explicitly states that he is and even gives strong indications that he isn't, such as him saying "but humans… you’re messy" (not "we're messy"). And his smile is "too wide," which might indicate failure to properly mimic empathy. So perhaps an android?
2. The man is a senior Empathy Calibration Technician, which is the one job still available to humans. It would be perfectly within reason to expect Elian to be tested by another human with more experience in the same role. (But then we'd have to ignore the hints about the man not being a human.)
Also, well spotted about 8 misreading #4's reference to "she" and misapplying it to Unit 734.
Finally, yes, the raindrops stuff is strange, especially since the same #8 passage circles back to sunlight. But we can generously interpret both rain and sunlight as anchors of the same emotional connection to nature and a bygone world.
I'll have more to say about some of your comments, but I'll hold off in case a few more people might want to guess.
Great subject! Im not reading other comments before leaving mine, I dont wan to be influenced. I belive you were number 8, though I think 4 or 5--cant remember which, also caught my eye. Gunna hit post then read others
I like the first section; I think you wrote the first section. It gets increasingly cringy ongoing but I wonder if there is also a factor of the diff models only writing a piece?
Yeah, the “handover” effect is definitely a factor here. I'm sure one of the better LLMs would've done a better job on its own than in this combo-writing experiment.
I also quite like the first section, but I won't reveal whether I wrote it just yet. Let the people guess!
Also, remember that my goal is to try to blend in, so liking a section might not necessarily mean it's mine.
Shared to other writers as was fascinated, fantastic story and great pitch rem Count Zero insight-level provocative
I vote section 8 human hand, the squeeze ball is confusing if not a human (why does a robot have a stress toy), I think the woman reference is too illogical for a machine
Stay tuned for the reveal next Thursday!
Just remembered you saying AI can be funny. This is a different subject but Chapgpt has given me some great stuff--look at this https://medium.com/@marvin.vonrenchler/breaking-news-chatgpt-is-getting-married-edb5813e4cdd
Yeah, mimicking an accent is right up ChatGPT's "parody" alley. I mentioned that wordplay, parody, and analogy are the easiest types of humor for AI here: https://www.whytryai.com/i/147880162/ai-excels-at
I'm pretty sure the first part was ChatGPT, based solely on the name "Elian". It's used that name in a sci-fi story for me recently, and I've noticed it tends to often use the same or very similar names across different stories.
I think 2 is AI as well. "Humans are messy" is the kind of thing AI's like to say. I once had two GPTs pretend they were a human administering a Turing test on the other, and it was full of them saying back and forth how "messy" and "real" and "raw" each other's responses were.
3 has to be AI. "Now, its grief isn’t sadness. It’s silence." Unless you included that as an intentional curve ball...
I'm much less confident for the rest, but if I had to guess, I'd say number 6 is you. I'm not sure exactly why. I think the number in 7 is probably a sign it's AI, but I'm not sure. 6 also seems to return to about the right section length, whereas the others had been getting gradually longer.
I enjoyed the whole story. I'm OK with a little slop, just like I'm OK with a little fast food.
I like using AI to do "Choose Your Own Adventure" stories, like I used to occasionally see as a kid. The kind where at the end of each page it would ask how you want to respond, and tell you what page to turn to for it. They were "slop" too, but it's fun and I don't care.
Nice! Concrete guesses backed up by experience - love it!
My very first attempt at making a Custom GPT was "Quest Weaver," which was a CYOA-style story maker. I also wanted to add RPG elements where people could create a character with attributes, and the story would be affected by those (and their decisions). Then DALL-E 3 would generate images for the scenes and locations. It didn't quite work at the time, as I described here: https://www.whytryai.com/p/build-gpts
But that was ages ago, during the early days of GPTs, so maybe things have improved.
I originally planned to only briefly reveal who wrote which parts of the story, but now that there are so many guesses and opinions, I'll probably expand it into a full post with some observations about AI fiction.
Stay tuned for next Thursday!
My guess is that 5 is the human. Just going off vibes. Also, it’s an incontrovertible fact that no elevator has a mirror.
Stay tuned for the reveal next week! And “elevators don't have mirrors” sounds like a plausible-sounding LLM hallucination in its own right.
This is one of my favorite posts of yours.
And we get to play a game!
I also think you wrote #1.
What is interesting is that whoever wrote #2 immediately deviated from the prompt: "In a future where ALL jobs have been fully automated..."
We then meet the, um, human man in a gray suit. 🤔
It's also amusing that #7 mentions SUN three times but #8 jumps straight into... storms and raindrops.
#8 I think also picks up on #4's female reference: "She'd told him as much. She was right." and makes the unit a woman.
I'm looking forward to the answers next week. 😁
Maybe we can also get a resolution to the story from the winning LLM?
Without hopefully spoiling too much, I'd like to point out that there's no conflict in #2's passage and the original premise. I see at least two possible interpretations here, under both of which the premise still holds:
1. The "man" in the suit isn't human. The text never explicitly states that he is and even gives strong indications that he isn't, such as him saying "but humans… you’re messy" (not "we're messy"). And his smile is "too wide," which might indicate failure to properly mimic empathy. So perhaps an android?
2. The man is a senior Empathy Calibration Technician, which is the one job still available to humans. It would be perfectly within reason to expect Elian to be tested by another human with more experience in the same role. (But then we'd have to ignore the hints about the man not being a human.)
Also, well spotted about 8 misreading #4's reference to "she" and misapplying it to Unit 734.
Finally, yes, the raindrops stuff is strange, especially since the same #8 passage circles back to sunlight. But we can generously interpret both rain and sunlight as anchors of the same emotional connection to nature and a bygone world.
I'll have more to say about some of your comments, but I'll hold off in case a few more people might want to guess.
Great subject! Im not reading other comments before leaving mine, I dont wan to be influenced. I belive you were number 8, though I think 4 or 5--cant remember which, also caught my eye. Gunna hit post then read others
Cool. Let me know which of your three guesses you settle on once you've read the full thing! I will reveal who wrote which part in next week's post.