Artificial intelligence should support and extend human creativity, according to a principle developed by National Gallery of Australia staff as the institution works out where the technology belongs in its operations.
For chief information officer Jade Carson, that means building the governance structures and the confidence to experiment, while keeping the gallery’s purpose at the centre of decisions about what to buy, develop or use.
In this episode of Delivery: A GovTech Podcast, Ms Carson speaks with James Riley about introducing AI to a cultural institution, the technology behind its exhibitions and the foundations needed to support new capabilities.
Ms Carson is also the gallery’s nominated accountable officer for artificial intelligence. Over 12 months of working with staff, she has helped develop principles covering accountability, transparency and sustainability – but the principle staff drew out themselves speaks directly to the institution’s mission.
“To ensure that we support and extend human creativity, and not replace it,” Ms Carson says.
The gallery’s executive and council support exploring AI for productivity and innovation. Ms Carson’s task is to give staff a safe environment in which to test these possibilities, with governance that supports them when they take the next step.
Some of the most promising applications could be behind the scenes. While attention naturally turns to artworks, Ms Carson sees opportunities in building operations, security, potentially conservation, and information and records management.
Discussions with curatorial teams have also identified research opportunities, where digitising historical information would give staff material they could draw on in future.
The gallery remains in a pilot and experimental phase. Business areas can pitch to the director for funding from a dedicated pool, explaining where they expect AI to deliver productivity improvements or innovation.
“We do have a restricted budget as a cultural institution, so we’re taking really small steps from a trialling perspective,” Ms Carson says.
She regularly compares approaches with CIOs across cultural institutions, including how limited trial budgets can help establish what useful applications actually cost.
For sensitive uses, the gallery is incorporating AI oversight into its existing governance arrangements, drawing on Digital Transformation Agency guidance.
“We’re not building anything separate, so we’re able to bring AI oversight into the governance processes we already have – including our existing ethics committee,” Ms Carson says.
She says AI uses involving First Nations collections would be treated as high risk under the DTA’s approach. At the gallery, such a proposal would be considered by a working group, then an AI ethics committee and council. At the time of this interview, that escalation process had not yet been used.
The wider technology task is equally tied to the institution’s particular needs. Alongside networks, cyber security and workplace systems, Ms Carson’s team supports exhibition technology, including digital artworks built on platforms chosen by artists.
One assignment even involved penetration testing a giant robot, an example of the unusual demands that artworks can make on the technology team.
This variety requires costs for specialist skills within a small team, and makes reliable underlying infrastructure essential.
Ms Carson is in the final year of a three-year strategy covering infrastructure, service management, information and records management, and cyber security. Addressing technical debt is part of the work, including removing barriers to automation.
The aim is to leave the gallery ready to support what staff ask for next, with AI among the opportunities made possible by those foundations.
For Ms Carson, the test remains whether a tool serves the organisation’s purpose. She points to other cultural institutions finding applications that directly support their core work, rather than experimenting without a clear need.
“It’s important to find the real use cases within an organisation and implement the technology to support them, rather than the other way around,” she says.
Do you know more? Contact James Riley via Email.