Anthony Harris, Research Fellow (Regent's Park), Visiting Fellow (Kellogg College) and British Academy Ker Award Holder
Although I have an industrial and academic background as a computer scientist, I am primarily a systems designer. Hence, I am able to visualise complex systems but implementing them has always been a challenge because I am not a natural coder and it always takes me time to produce working solutions. However, since last year I have been using AI Coding systems and generative AI to implement my ideas/visions to great effect. 'What I can imagine, I can build 'at speed'. The only limit to what I can do with generative AI and AI coding agents has become my own imagination.' (Harris)
I have an MA in English Language and Literature from Oxford, a first-class MA Res in Medieval Studies, and a humanities PhD from Cambridge (early mathematics and astronomy) where I used digital humanities to great effect. I am a specialist in early manuscripts, and these are normally described in what are known as 'summary catalogues' (SCs). Many of these SCs are now online as PDFs but OCR (Optical Character Read) quality is often poor and free text search is limited to what is available through PDF searching. I have been using OpenAI’s Codex coding assistant to generate Python programs that use the OpenAI GPT-5.x engine to re-OCR these PDFs, analyse their structure, infer aspects of genre, value, type etc., and export the manuscript records to Excel. Once in Excel I can then use the data to sort by date, size, genre, classification and also export to XML (TEI).
In previous years I have tried to do this many times using Adobe Acrobat and other PDF tools to export summary catalogues to spreadsheets. However, either the OCR has not been good enough or the Excel export functions have been inadequate. The latest OpenAI models, when accessed through the APIs, can not only OCR the pages but can also infer data structures which means that the extracted data is much more useful for research and analysis.
My advice: be adventurous. 'The only limit to using generative AI is your own imagination' (Dr Anthony Harris, 'Letters', 'The Times' newspaper, Saturday May 17 2025).
Should you wish to discuss this application with Dr Harris directly, you can reach out to him at tony.harris@regents.ox.ac.uk
Mapping geographical bias with GenAI
User Case Study