Marieke van Erp

Reflections on Reflection in Humanities and AI

The fields of Computer Science and Humanities operate in different ways: different research methods, different publication cultures, and different perspectives. But with interdisciplinary collaborations, projects, and teams, the use of computational methods is becoming more widespread in humanities use cases. As many archives have been and are still being digitised, automatic analysis methods are a necessity. However, historical texts present computational analyses with many different challenges such as digitisation artefacts, segmentation, language evolution, and changing societal values and large generic models are often not optimised for this domain resulting in suboptimal performance. In this talk, I will argue that combining the skills of humanities scholars with AI is beneficial to both fields: humanities scholars can benefit from new tooling to deal with big data in their domain, and AI researchers can benefit from the humanities’ long tradition of source and tool criticism.