Fun project. The AI services only took around 3 hours total to build and tweak. The UI took probably another 6. But its amazing what someone can do these days...
Made a fun AI app. It generates "Stories(a la snap chat)" for children that teaches them about topics.
All real generative responses, the information(GPT) as well as supplementary graphics(Stable Diffusion). GPT will also suggest follow up topics for less typing.
Its not fast(as you see), its not cheap. 2¢ or so a slide.
But it could be replaced with google image search, or a cached image database. But i preferred Stable Diffusion over Unsplash API almost every time.
@CaseyNewton@ZoeSchiffer March 22, tweets disappeared from timelines and appeared only after the user refreshed the page.
April 1, all tweets inexplicably changed into images of cats and dogs.
April 8, users spontaneously began tweeting gibberish.
This last one seems a little farfetched to me.
@CaseyNewton@ZoeSchiffer I let GPT continue your timeline:
On March 5, Twitter was down for several hours and showed an error page.
On March 10, all posts were wiped from the timeline.
On March 15, Twitter transformed into a video game where users had to collect tweets to complete levels.
@jonathanhammond Made this fun one a little while back. You can input any story idea and it makes a short book with (somewhat inconsistent art) right at my kid's reading level. We've been using this every night to read!
GPT-3 + DALL-E, and some prompt engineering