Mo von Bychelburg was awarded a bursary to attend the Digital Humanities Oxford Summer School in 2024. To join the mailing list and learn about the next summer school sign up here. Read about Mo's experience at the summer school here:
This summer, I attended the DHOxSS 2024 Summer School at Oxford. I am a PhD student in Information Studies at Uppsala University, doing research on AI in cultural heritage archives, and was recommended this summer school by a colleague of mine. Actually, my colleague made this summer school sound so interesting that I chose it over another summer school that I was technically required to attend! Of course, the coordinators of the other summer school understood my decision, and yes: Choosing DHOxSS was definitely worth it. Despite “only” attending online, and not being able to visit the beautiful city of Oxford, I am glad I chose the summer school this year. I learned a lot of interesting things about areas of AI that I am not familiar with.
On day 1, David de Roure explained to us that he considers creating music with AI as co-creation; the human is the curator who uses the AI. In the next lecture, Dominik Lukes talked about something which I consider the most important “take-away” for my own work: He explained the vectors involved in Large Language Model technology, stating that “AI predicts the next word by geometry”. I had tried to wrap my head around the use of vectors/geometry in LLM, but Dominik managed to explain it in a way that it finally made sense to me! He also gave us a list of AI tools to try out, such as claude.ai, elicit.com, notebooklm.google, and consensus.app.
On day 2, David introduced us to some intriguing concepts and projects. For example, he talked about community collaboration projects such as Wikipedia and Galaxy Zoo, where volunteers and scientists work together to build a knowledge base. Next, Matthew Nicolls showed us how to build models using the program SketchUp. He talked about what questions to ask oneself when building models, such as: What is the date of the model? Who/what is the model built for? What do I need to make it work? Then, Jack Orchard introduced us to gaming and AI. I remember an interesting comment from a listener in the chat: “Connecting to LLMs, there is a lot of work going on with NPC and how AI can improve the, mostly, scripted interactions”. Poppy Wilde discussed her experience with online communities and the importance of having a critical lens on your own role as a researcher. Finally, Leighton Evans talked about XR (extended reality), immersion and digital twins. Importantly for me, she commented that, as digital objects can disappear in the event of a technical failure, archives are important!
On day 3, Alwyn Collinson discussed AI-generated texts, mentioning the fact that “well-written texts” had long been considered “as a sign of deep engagement with a topic”, but that this changes with AI. Terhi Nurmikko-Fuller and Matti Vuorre again took up the topic of games; with Zakiya Leeming, we learned more about music and AI. Megan Gooch talked about AI for text recognition, discussing such topics as metadata, LLMs and AI “hallucinations” – which according to her would be more accurately termed “bullshit”… Nanna Bonde Thylstrup then discussed with us the incredibly large amount of data we have amassed, as well as preservation/archiving and censorship issues.
All in all, DHOxSS was a very rewarding experience, and I want to thank all the coordinators and speakers for creating this event for us. I can really recommend everyone working with Digital Humanities to join this amazing summer school! I would be happy to join again next year.
