Quality Assurance of Application of AI and Machine Learning on Art and Cultural Heritage – Opportunities and Challenges for Future Art History

Abstract

AI advancements enable new art-historical research tools, like OpenAI’s GPT-4 image-to-text function. Emerging cross-modal translation models can link texts, images, and more. Quality assurance in digitising art sources and training AI models is crucial. Concerns include source origin, content curation, presupposed image models, and potential cultural bias. It is essential for art historians to influence digitisation and AI data processes to ensure quality research when using AI-based tools.