Diella was appointed as ‘minister of state for artificial intelligence’ in Albania late last year, with the aim of guiding government work to curb endemic corruption in public procurement. The minister is, however, not human. She is AI-generated.
Is this a symbol of futuristic governance, an interesting step forward, or just a gimmick?
“I think it’s thought-provoking and promotes discussion,” Sheena Thompson, global government and infrastructure AI leader at EY, tells PF. “We have so far largely seen government departments using AI in the back-office. Citizen-facing usage has been quite restricted. Showcasing the use of AI to drive decision-making is shifting the ambition level quite drastically.”
Diella began as a tool on Albania’s online government portal, used to help citizens navigate services. Since September, responsibilities for procurement have been gradually transferred to the system. The stated aim, according to Albanian prime minister Edi Rama, is that “public tenders will be 100% free of corruption”.
“It is a prime use case,” says Thompson’s colleague Thomas Erwin, EY’s global co-lead for AI in government and infrastructure.
He continues: “Promoting ‘her’ to minister is partly provocative – a political statement given the situation with corruption in procurement in Albania. Diella doesn’t really have agency. It’s a decision support system for humans. In that sense, behind the politics, it’s very much in line with what we are seeing elsewhere in the world.”
Selling Diella to the public is not so much a question of agency, then – human beings still make the final decisions – but rather perceptions of agency. “The question is: to what extent in the public space are we willing to transfer agency from humans to technology?” Erwin says.
Citizens can take some convincing. “It depends on the country and demographic,” Erwin explains. “My generation and older are very sceptical – although they’re willing to use any app on our phones and give away our most private information, they can be unwilling to give away even one piece of information to the state. Whereas younger people are more than happy to trade ease of interaction for giving away parts of their data.”
Thompson explains: “When it comes to trust, people want to know how the AI tool works: what’s inside the black box? To what extent is it using logic and applying it to decisions? I do think there’s an oversensitivity to that; if it was a human making those decisions, we wouldn’t know what was inside their ‘black box’ either. And actually, one of the drivers of Diella was to focus on areas where we actually want to remove human discretion.”
“Change management is the age-old problem,” Erwin says. “A particular thing to be on top of with AI, though, is the speed with which this technology is maturing and evolving. The progress is almost on an exponential curve. And people aren’t always great with incremental change, let alone exponential.
“One of the biggest challenges we are discussing with governments is the private sector experience and the citizen experience are already diverging.”