Joe Nockels
Projects
01
Terms and Conditions: Interpreting AI Tool Licences in Research Contexts
University of Sheffield, University of Birmingham, University of Buffalo, Programming Historian
At present, university libraries are limited in recommending AI tools to students due to certain terms and conditions. These conditions, in turn, lead researchers to accept individual risk through unauthorised 'shadow' usage. In modelling typical red flag clauses in AI licences through Natural Language Processing, this project outlines recommendations to tool providers for better incorporating library needs and their situational AI use.
02
Recognising Text, Recognising Processes -Explainable Automated Text Recognition for Scottish Spiritualist Newspapers (2025 - 2026)
University of Sheffield, National Library of Scotland
This project researched how Explainable AI (XAI) developmental principles intersect with library requirements, and can support sustainable digital transcription at-scale. Directed by libraries' growing emphasis on XAI, this project answered: what aspects of ATR need further explanation? How are developers positioning themselves toward XAI public-facing documentation? Can XAI increase library audiences’ trust in ATR resources? Does greater explainability compromise ATR performance?
03
Using Handwritten Text Recognition (HTR) on The Life and Times of Frederick Douglass (1846-1894) (2024)
University of Edinburgh, Library of Congress
This project demonstrated the efficacy of text recognition to analysis pipelines on handwritten library collections, using the autobiographical works of Frederick Douglass (1818 - 1895). Following this, it attempted to understand how intimate histories contained in handwriting can challenge and develop cultural understandings previously reliant on mechanical printed text.
04
.png)
Making the Past Readable: The Impact of Handwritten Text Recognition on Libraries and Their Users (2020 - 2024)
University of Edinburgh, University of Glasgow
This doctoral study considered the socio-technical infrastructure surrounding the AI-enabled transcription of library collections. Despite sectoral awareness, institutional approaches to HTR still display gaps between conceptualising and operationalising workflows. Therefore, this project assessed the impact of HTR on libraries, users, and the wider information environment. Such analysis resulted in a set of recommendations for HTR’s future provision, directed thematically at libraries, users and developers.