I wrote about trying out LLM models for document transcription, inspired by @benwbrum (and Sara!)
https://cdrhdev.unl.edu/log/2025/comparing-ai-models-for-document-transcription/
Discussion
Loading...
@nirak @benwbrum Have you considered setting this up as a #citizenscience project on a platform like the Zooniverse? They have some transcription projects going on, e.g.:
https://www.zooniverse.org/projects?discipline=history&page=1&status=live
@nacly I can't speak for @nirak, but while I like the Zooniverse and am friends with some folks there, I run my own platform at https://fromthepage.com/ where we recently added Gemini support for creating "AI Drafts" which transcribers may use (if they wish).
@benwbrum @nirak Wow, TIL FromThePage.com so interesting, thanks!
Have you published any lessons learned, e.g. on community governance, relationship with volunteers, analyses on why people volunteer, etc. for the platform? The Zooniverse have shared lessons like that and I'd love to compare/contrast that experience with FromThePage!
@nacly We've published a lot on our blog https://content.fromthepage.com/ and our newsletters https://content.fromthepage.com/our-crowdsourcing-ai-newsletters/ and host monthly webinars which are recorded at https://www.youtube.com/fromthepage
Our origins are a bit different from Zooniverse, in that we started out doing full-page transcription and later moved to structured data, while their journey has been in the other direction. That changes some aspects of the platform--e.g. volunteers can choose their work and navigate to pages.