This was so much fun! I still find myself laughing when I remember @Gabriel_Alejandro’s hair flip
Thanks for organizing! This was fun.
Seattle AI Society meetup group members watching an AI script-reading performance by other off-image group members.
Seattle AI Society members reading an AI-generated script based on prompts they provided to ChatGPT. (Note: I have a couple more pics, but I’m limited as to how many I can post until someone else posts/replies).
Coincidence between the use of GPT4 at this meetup and some feedback in the community around how not-useful GPT4 is for anything creative. One example: https://twitter.com/KevinAFischer/status/1637505023142944770
After 48 hours with GPT4 I can safely say people don’t realize how problematic its censorship is.
I deeply respect the team but the unfortunate truth is that GPT4 is borderline useless for anything other than productivity applications.
And that’s not the worst
Whether it’s fiction, script writing, video game design, therapy, etc
If the action gets at the heart of human experience their application of RLHF has at a minimum made it much harder to express your project, and in many cases literally impossible
Look at that narrator!
I see a villian back there!
I don’t get the connection between the quality of AI-generated fiction/script writing, etc. and censorship. I feel like there is a lot of context missing. I guess I can see how if you remove all car chases from an AI’s training data, then it probably won’t ever write a car chase scene? Or if you train it on data which includes people in black hats, but you restrict the AI from outputting any mention of people in black hats, then the villains in cowboy movies it writes will be wearing some other color hat?
Members of the Seattle AI Society meetup group discuss the outcomes of the ChatGPT scriptwriting event.