I am a hacker looking for a hacker to cobuild an awesome side project.
I’m learning GPT-3 API/docs this weekend, and you should HMU if you’ve built something on GPT-3 already (or plan to and have the skills). I also have experience with APIs, and ML/DL models. (creds)
I’ve seen what the current GPT-3 models can do and I’ve played on the Playground (asking it some questions that were decent). It’s been done before and others have posted about it in this forum, so nothing new, but I think I can leverage the API to make a much better script. My brother works in Hollywood and writes scripts professionally, and he’s helping me. (his stuff)
It’s clear GPT-3 can write scripts, and those scripts have been filmed… great work but so far I didn’t enjoy any of them.
I want to use the API and write a story I can slap a human name on and give it blind to a successful producer that actually kinda likes it. Well… they at least must enjoy having taken the 20 mins to read it or watch it, otherwise it fails the Content Turing Test (a twist on the classic test).
It could be, but I haven’t read/viewed any of them yet…
What I meant by saying the “Content Turing Test” is a slight variation on the original, as related to content, which must be enjoyed to pass the bar, and it’s less important thinking it came from a human. It’s not enough for GPT-3 to simply look like it can write English, it must write something people choose to - and gain value from - dedicated their time to. (and obviously if they pay for more)
But because I believe humans may judge a work less enjoyable if they know in advance it was AI-generated, I would simply cover that up with a pseudonym. This is because my goal is to generate enjoyable content, not build the first AI to pass the Turing Test. I think Google already did that almost, anyways.
Well after chatting with some people here and in the Startup School Founder Match program, some ideas did come up. My end goal is everything, everywhere, from animated shorts to Blockbuster action movies to Bollywood to fancy French art house films… and everything in between. But that’s some years out.
A great “MVP” idea was to get a beachhead starting with animated kids shows. Obviously the first few generations of the system would not pump out scripts ready to win Best Picture, but kids are a less discernible audience. I’ve watched some of these shows with my nieces/nephews and they honestly sound like they were written by a robot already and incredibly formulaic, and even the animation quality can be low, yet the kids come back clamoring for more.
So I think starting with short, English language, animated kids shows is a great type/genre to start with… but as I said, eventually all types of movies in all languages and genres is the goal.
and incredibly formulaic, and even the animation quality can be low, yet the kids come back clamoring for more.