Understanding responsible use of AI in decolonial practice

What are the tensions between AI use by museums and their decolonial practice – and how might these tensions be navigated?

These questions guided Coloniality, Collections and the Responsible Use of AI in Museums and Heritage, an AHRC/BRAID-funded collaborative project by Sheffield University, Cambridge University, Sheffield Hallam University, York University and the Royal Armouries Museum in Leeds, the findings of which are now published in a toolkit and workbook.

Designed with decolonial values in mind, the aim of the toolkit and workbook is to support museums and heritage organisations, those who work in them and their broader communities, to define on an ongoing basis what a responsible use of AI should look like for their specific setting, ambitions, and projects.

AI is enmeshed with and maintains historic and ongoing colonial power relations, biases and harms. Given the momentum around AI integration, museums and heritage institutions face a complicated and potentially fraught process of navigation in which they attempt to reconcile AI use with their educative, social and cultural mission and ethics – including their processes of decolonisation.

To explore this, the project team used the Indian collection at the Royal Armouries in Leeds as a case study. They worked with museum staff and consulted extensively with museum and heritage professionals across the UK, including at a workshop held in summer 2024.

By focusing on the Indian collection, the team were able to explore the ethics and politics of the data that museums hold on objects acquired in the context of colonialism and the potential implications for decolonial practice of using this data within AI models.

In particular, the team focused on the museum's catalogue data and explored the implications of using this with an AI model. Analysing the collection and catalogue from various perspectives, the team used speculative design to explore the ethics and politics of different AI use cases.

Central to the project's approach was bringing interdisciplinarity to bear on the ethics and politics of museum AI use and the question of how this might be navigated in practical terms.

The team's disciplinary backgrounds spanned politics, oral history, computer science, English, and digital culture and art, and experience working in and with the museum sector.

Throughout the project the team's approach was to work in dialogue both with museum sector approaches to decolonisation, including the Museums Association guidance on decolonisation, and with ideas and debates about what “responsible AI” can and should look like.

The resulting toolkit is divided into sections that provide background and useful definitions and perspectives related to the tension between AI use by museums and heritage institutions and their decolonial practice.

The final section sketches a pathway for critically engaging with AI and developing situated approaches to responsible AI use. Links to relevant resources are provided throughout the sections.

These resources have been chosen to support an ongoing programme of work and the development of connections and relationships to underpin a person, place, and culture-based view of what responsible AI use can and should look like. Each part of the toolkit is accompanied by opportunities for solo or collective reflection or exploration in a workbook format. These are designed to help readers to begin to apply and think through the ideas in relation to their own organisation. 

One of the key messages of the resource is that responsible AI is not a panacea and a “navigation” of AI ethics should not assume that some form of AI use will result.

Deciding not to use AI, whether in a specific context or across an organisation, is a valid response to the ethical problems posed by AI. Not using AI should be on the table in any discussion about what is right for a setting, and opting out of AI should be safeguarded as a viable path for individuals and organisations.

• Access the toolkit and workbook

Joanna Tidy is a senior lecturer in politics at the University of Sheffield