By Gabriella Giannachi

In my last blog I talked about the importance of documentation for artists, museums, archives, and computer scientists, establishing the reasons for documentation. Here I look specifically into how to document artworks and what benefits and values may derive from documentation. Historically, various disciplines have approached documentation with different agendas and through different methods. Thus, in performance studies, the tendency has been to focus on the relationship between documentation and performance from an ontological or a phenomenological perspective. This means that the debate has concentrated on whether a performance and its documentation may coincide, and/or whether documentation may generate a performative experience. In art history, however, the two have been looked at relationally, and, since the introduction of performance in museums, documentation has often been used generatively to activate new iterations of artworks.

Organisations in the cultural sector developed varied documentation guidelines and frameworks. In view of the fact that small museums and organisations often have limited resources for documentation, the AHRC-funded Documenting Digital Art project (2019-23) developed a best practice documentation framework that aimed to help artists, audiences, cultural organisations, and policymakers to navigate the field. This includes a step-by-step guide which highlights the values of documentation, including its ability to provide a snapshot in the life of a work of art, for various aims, such as conservation (and so also activation), presentation, and dissemination. The framework talks of the values brought on by documenting responsibly; the importance of including different stakeholders; the use of open systems; the values in recording the context in which a documentation has taken place; to highlight the interconnectivity among different types of documentation; to identify risks from the outset; and remember the role played by temporality, highlighting the advantages of pursuing decentralised rather than corrective versions of documentation.

A subsequent paper, ‘Using AI in documentation’, written with Annet Dekker, which investigated the use of ChatGPT for the documentation of artworks by Bay Area artist Lynn Hershman Leeson, illustrated that AI’s ability to document artworks is commensurate to how much information already exists online about an artwork. Having said this, it was found that AI could provide additional contextualisation; link artworks across different collections; and generate prompts for novel audience experiences. AI may thus facilitate processes related to extensive and heterogeneous collections, proposing methods for arrangement and classification, as well as generative creative prompting.

The use of AI in the Somabotics project occurs at different stages of the creative process and through different models. Thus, Embrace Angels, developed by the Amsterdam-based duo and science researchers Lancel/Maat (Karen Lancel and Hermen Maat), aims to investigate how humans and robots can embrace each other. More specifically, the artwork explores the social meaning of touch, looking into how robots can consent to and mediate touch. In her review of a Somabotics panel about Embrace Angels offered at LI-MA’s Transformation Digital Art Symposium in March 2025, the art historian Nina Knaack pointed out that Somabotics’s research for Embrace Angels shows that AI documentation not only functions as a conservation or storage mechanism, but that it also interprets, translates, and even hallucinates new representations of the artwork.

Whether carried out by the artists, or the team, or by both, as shown in the image documenting Hermen Maat documenting Karen Lancel’s robotic embrace, the use of AI in the documentation of Embrace Angels sits both outside and within the artwork. Hence, the challenges in documenting this kind of work without being too prescriptive and maintaining a decentralised view in the end may also be a challenge about what to capture so as to produce a public facing archive that doesn’t consist of hundreds of unrelated documents and yet tells the (his) story of this generative, participatory, performative, hybrid, somabotic work from multiple perspectives.