Rhetoric on AI is rarely backed up by resources – panellists

Although central banks talked a great deal about leveraging artificial intelligence, they were less willing to allocate sufficient resources, a senior official from an advanced economy institution told a discussion at the Central Banking Summer Meetings.

At the panel event in London on June 12, an official from a second advanced economy central bank added that innovation was “crucial for smaller central banks. We need space to fail. But we learn from those failures”.

However, a panellist from a different region said resource-related hurdles tended to disappear when the results were positive.

“We might have a vision for the future,” they said, “but we still have to deal with a lot of obsolescence. Budgets are prioritised to fix things in the short term, while we keep an eye on the long term.”

Another panellist said not all projects they had worked on in the realm of innovation would come into existence for public use. Departments for innovation were not “production shops”, they said, adding that the goal was to identify gaps where technology could be leveraged to improve banking operations.

The official from a second advanced economy central bank agreed that “if you experiment with technology, the benefits are incremental”.

The panellists also considered the potentially malign role AI could play, especially with regard to financial stability.

One of the panellists outlined three core pillars for the use of technology in central banking: aligning data to regulatory outcomes; ensuring metadata was actionable; and maintaining observability and ethical governance.

“Technology is not the end of the game,” they warned. “Sometimes it’s just good to have a tent while you are building a castle.”

Speakers acknowledged that legacy challenges, fragmented systems and limited budgets often hindered regulators’ ability to act on promising innovations.

Panellists praised the role of global innovation hubs – such as those led by the Bank for International Settlements – in offering blueprints, code and research that regulators around the world could adapt.

They cited the BIS Innovation Hub’s Project Nexus as an example of a project that successfully transitioned into real-world implementation, linking national payment systems across borders.

Yet panellists were candid about the reality, and recognised that only a fraction of experimental projects reached the production stage. For most, the value lay in learning and knowledge-sharing.

“It’s not always about scaling a solution,” one speaker noted. “Sometimes it’s about sharing the problem well enough that others can build from it.”