Realising the potential of AI in museums

Jonathan Knott, 14.02.2017
How can cultural institutions make the most of AI technology?
Museums will need to change traditional ways of working and conventional thought processes to fully embrace the potential of artificial intelligence (AI). Here are five key ways museums can achieve this.
Build partnerships

Most expertise in AI lies outside the museum sector, so institutions that want to benefit from it will need to look beyond their walls.

Andrew Larking, the creative director at digital agency Deeson, says: “What needs to happen is for this sector to build relationships with the thinkers and practitioners who are at the cutting edge, and to put in place a programme of works to test and study the outcomes that AI could create.”

“You can't think of AI like you do about customer relationship management or a website.  The skills [required] aren't as directly transferable across sectors or departments,” he adds.

“It's like introducing design practice into a technology company – it happens over years, not overnight.”

Consider the broader implications

AI raises fundamental ethical and social questions. A British Library event on machine learning planned for September will consider what implications AI has for wider society, as well its usefulness for the museum sector.
The event will look at the use of machine learning in crime and policing, as well as politics and electioneering. It will also consider broader ethical questions such as where the responsibility for the results of machine learning lies, and how these systems can be monitored for fairness.
“The aim of the event is to get experts together to increase people's understanding of the wider implications of machine learning and digital technologies on society,” says Mia Ridge, the digital curator at the British Library.
Ceding control

Because AI is an experimental area, it’s not always possible to know what the results of projects will be, and, therefore it’s important for museums to be willing to experiment and keep an open mind.

Mario Klingemann, a Munich-based digital artist who conceived the X Degrees of Separation project as artist-in-residence at the Google Cultural Institute (link to feature), says that the experiments at the institute’s ‘lab’ in Paris are not designed with a specific purpose.

“Not everything is super-useful at first sight, it might be playful – the purpose of a lab is to try out different things,” Klingemann says.

According to Larking, the ability to relinquish control over content is crucial if museums are to get the most from AI curatorially.

“If content teams were prepared to stop having total control over content, and instead could learn to define and write the constraints and success criteria that AI systems use, you'd start to see output at a quality and pace that human teams won't be able to compete with,” he says.

Fostering collaboration

A recent piece of research by the innovation charity Nesta used machine learning to estimate the size of museum visitor numbers using social media data.

In a blog post about the project, Nesta research fellows John Davies and Antonio Lima say that the government could help museums to develop their use of new data sources, as they look to overcome risks and capacity challenges.

Davies explains that in this fledgling field, fostering collaboration within the sector is an important issue, although he is cautious about saying exactly how the government could help.

“Because we are dealing with quite new techniques and diverse sets of organisations, one wonders how knowledge can be most efficiently spread through the sector,” he says.

Laying the right data foundations

Brendan Ciecko, the chief executive of Cuseum, a digital platform for cultural institutions, says that museums’ use of AI with their collections is “still quite nascent”.

Getting the most out of the technology will depend on establishing the right open data formats and standards. “For most institutions, there is still a lot of early foundational work on the collections data that would need to take place,” he says.

He believes that large organisations like London’s Tate, the J Paul Getty Museum (Los Angeles, US) and the Museum of Modern Art (New York, US) may take the lead in developing uses of AI, “given the quality of structured data and the resources that they have”.