Use a multimodal AI api to extract questions and answers from the newspaper and display in text instead of via the image

danielemery / quizlord-api

Graphql api for sharing newspaper quizzes between friends, including results and statistics

https://quizlord.net

MIT License

1 stars 0 forks source link

Use a multimodal AI api to extract questions and answers from the newspaper and display in text instead of via the image #93

Open danielemery opened 6 months ago

danielemery commented 6 months ago

Since multimodal AI is getting really good, it should be trivial to extract the individual questions and answers for quizzes to provide a better user experience.

Add support for questions and answers to be stored in the database
Implement queueing system to run a task whenever a quiz is uploaded
Display questions and answers in text (leave the image as a fallback)

danielemery commented 5 months ago

Had issues with Gemini due to it not being available in Germany at this time. Going to look into Claude as an alternative https://github.com/anthropics/anthropic-sdk-typescript