The Impact of Generative AI on the Digital Humanities: Disruption in Research and Education

ORA (Ownership, Rights, and Attribution) is a novel technical framework for media tokenisation which uses distributed ledger technologies (DLTs), smart contracts, non-fungible tokens (NFTs) and the C2PA standard for content authenticity to explore alternative media ecosystems. ORAgen is a web application that demonstrates the technologies underlining the ORA framework by allowing users to create, tokenise, licence, and remix simply colour collages. Our ongoing research at DECaDE research centre uses this online demonstrator to understand how the emergent technical features, qualities, and constraints of media tokenisation, intersect with the real needs and challenges that professionals in cultural and creative industries (CCI) face, and traditional legal and cultural interpretations of ownership, rights, and attribution. This talk will highlight how this demo has been employed to reveal the value and challenges of attribution across the cultural and creative industries, and how this might offer insight into the growing interest and usage of generative AI in these fields. 

The training of machine learning systems on existing data sets brings great power to research operations through the scale at which this can be done, but it is well-documented that the imperfections and inequalities that exist in data sets can be echoed in the outputs, bringing problematic consequences, such as exclusions and clumsy (mis)representations. Teaching such issues is often impactful when utilising case studies of marginalised people, or little-known phenomena. However, by immersing students in processes of model training that involve themselves, their experiences and their communities, powerful insights can be generated that illuminate the workings of machine learning to students in subjects which critique, rather than create, the technology, and inspire them to produce creative responses. This talk will introduce observations from immersive learning experiences in media studies students, using accessible machine learning tools such as the Teachable Machine. 

Preserving email collections opens up vast research possibilities across various disciplines, enabling studies of individuals and organizations in unprecedented ways. It also presents significant challenges for archivists, particularly in reviewing emails for sensitive content that must be restricted due to legal or ethical reasons. Some types of sensitive content are clearly defined by laws and regulations, such as the General Data Protection Regulation (GDPR) in the EU, the Family Educational Rights and Privacy Act (FERPA) in the US, and the Health Insurance Portability and Accountability Act of 1996 (HIPAA) in the US. Other sensitive information may be more subjective and less well-defined. Currently, we use regular expression searches to identify structured information such as national IDs, phone numbers, and bank account details. We also generate keywords like 'performance review,' 'academic probation,' 'surgery,' 'litigation,' and 'marijuana' to surface potentially sensitive information. However, ensuring the comprehensiveness of these keywords is challenging. We are exploring two potential uses of Generative AI in this context. The first involves using Generative AI to generate keywords for searching sensitive information. The second use is employing the chat feature in Generative AI to help identify sensitive information.

AI has the potential to severely disrupt how we conduct research and education. Whether the disruption is positive or negative depends on us. One thing we cannot afford to do is ignore it, as this powerful technology is here to stay. In this talk, I will briefly explore some potentially ground-breaking applications of AI in the domains of research and education, both within academia but also beyond it (e.g. GLAM institutions). I will argue that there is at present quite a bit of hype surrounding AI, particularly the approach known as ‘generative AI’. Truly radical transformations of the way we carry out research and impart education will have to wait for more mature versions of the technology, especially those that successfully marry together symbolic and neural processing, but also for the maturation of other technologies that enable the collection of more fine-grained data. 

Large Generative AI in creative arts are shaping the creation and consumption of content, raising some old as well as some new questions at the forefront of EU regulation and governance. This talk provides a statement how framing Large Generative AI in creative arts for policy requires a multidisciplinary approach to the design, use and regulation of these technologies. It looks at the opportunities for artists, creators and fashion designers to employ a creative critique of AI and how a socio-legal framework should feed into the responsible usage of these technologies.