Reflecting on artificial intelligence and archaeology: the ArchAIDE perspective

The ArchAIDE project realised an AI-based application to recognise archaeological pottery, developing two deep learning algorithms to propose identifications based on images captured on-site while retaining key decision points necessary to create trusted results. One method relies on the shape of a potsherd; the other on decorative features. Developing the project meant facing challenges related to real-world archaeological data, deep learning techniques, ethics, epistemology, and hermeneutics. The project is still alive and moving towards long-term sustainability, which involves new challenges.