Acknowledgements
This exhibit is not only a result of technical engineering but also of a shared community effort to make AI-driven transcription more transparent and scholarly.
The concept of "Chain of Process" reasoning revealing the internal logic of a model to validate its transcription was deeply influenced by the insights and pioneering work of the following individuals:
- Ben and Sara Brumfield (Brumfield Labs): Pioneers in the world of collaborative transcription. Their work with FromThePage and their advocacy for IIIF standards have set the bar for how digital archives should engage with both human volunteers and automated systems. Their insights into how a "human-in-the-loop" (or an "AI-mirroring-a-human") should function and their experiments with this technology first let me explore these models.
- Dan Cohen: Whose recent article helped me understand that the model was behaving much like a trained paleographer.
- Mark Humphries: Whose recent article convinced us all this was worth exploring.
Technical Support
A special thanks to Mat Jordan for his reimagination and redevelopment of the latest version of Canopy which this exhibit is built upon.
About the Collection
The letters featured in this exhibit are provided by the Texas A&M University Archives.