#RSAA25 - AI-assisted Humanities Researcher Workbenches

The significance of generative AI for humanities research methods is poorly understood, and greatly underestimated—generally limited to anecdotes about where it makes mistakes and how students are using it to cheat. The affordances are quite astonishing. Where AI can generate near perfect Sanskrit, translate and analyse grammar expertly then research practice has changed radically. Where AI acts as a master’s level assistant able to instantly research, collate, analyse, and present, then then the practice of a digital humanist has changed radically.The methodological challenge is what to ask, how to ask it, and how to validate responses. As humanities researchers, we are ground truth. We need to engage with generative AI is as active collaborators rather than passive consumers—building workflows that accept or reject outputs, embedding emendations back into model training to align with scholarly standards. This presentation will explore how generative AI is being integrated across a range of humanities research workbenches we support: -Glycerine, a IIIF annotation workbench: has integrated an open-source Image AI and IIIF annotation pipeline, supporting iterative, scalable workflows for training models in image segmentation, captioning, and semantic tagging. -TLCMap, a workbench for mapping history and culture: has implemented an open-source mapping pipeline to extract and geolocate Australian place names from large texts. Researchers can review, emend, and validate results. -Omeka S, a graph-based content management system: we are designing modules that embed writing, translation, annotation, and visualisation tasks—powered by generative AI—directly into researcher workflows.