I haven’t found ChatGPT’s composition to be all that compelling or well-done. I mean, it’s impressive that the technology can do it at all, but does anyone have an example of anything like a 3k word story that is well-written and fun to read with a unique voice? Not saying this could never happen (with future models), but every time I’ve prompted a story it’s pretty weak and obviously AI-generated. What am I missing? #ChatGPT#WritingCommunity#fiction#AI#MachineLearning#LLM
@TimHoelscherX You’re pretty much spot on. It’s really only good at compiling amalgamated prose and flattening it out into something only somewhat coherent. It has little to no memory and can’t really come up with anything on its own. I only use it occasionally as a novelty tool or as a sounding board for ideas. Compelling prose is still human only for now.
@schizanon true and fair. I just wonder if people passing off LLM AI generated stories as their own work are pranksters or just delusional about how proficient this generated fiction is.
@TimHoelscherX probably depends how they prompt; chatGPT can take a lot of direction. You could feed it prose in a style you wanted it to emulate, but then you are taking a creative role in the process. Then there's always room for editorializing and proofreading. It's nothing but fancy autocomplete; it doesn't do anything on its own.
@TimHoelscherX@schizanon To me it seems like the same group as those who cheat on tests and buy their master's thesis from a ghostwriter... A combination of ambition, laziness, insecurity, and hubris. Which (maybe?) will always be there as long as art is a commodified product (which incentivizes shortcutting and "cost-effectiveness").
Similarly, online short-fiction magazines have been inundated with shit to the point that some have had to stop taking submissions from humans and robots alike.
@TimHoelscherX was wondering what motivates the subset of people who submit un- or loghtly edited AI output to publishers under their own names in the (admittedly presumed) assumption that no one will notice. I was generalising about them.
@TimHoelscherX ha, it works well for entertaining 7 year olds. imo, the people who say they're getting results this good are lying mfers trying to build a "following" as an AI influencer. It's good for certain targeted use cases where you can insulate it from consequences of failure, but nobody can reliably get it to perform like this (although GPT-4 is indeed a big improvement)
@kellogh it is an improvement, and I agree with you about the results in general. I know some pubs were getting flooded with generated story submissions and I’m thinking, “Whaaa?!” I thought I had to be missing something.
Add comment