Stay informed with free updates
Simply sign up to the Artificial intelligence myFT Digest — delivered directly to your inbox.
Central banks urgently need to “raise their game” to tackle the challenges and opportunities of artificial intelligence, as it transforms economies and the financial system, according to the Bank for International Settlements.
The BIS conclusions, outlined in a report released on Tuesday, underline the awareness of global financial authorities that they need to keep pace with the wave of innovation being released by generative AI, including large language models such as ChatGPT.
The organisation, which operates banking services for the world’s central banks, has carried out several experiments using the technology. It said AI was likely to be “a game changer for many activities and have a profound impact” on the broader economy and financial system.
“There is an urgent need for central banks to raise their game,” it added.
“Recent evidence suggests that AI directly raises productivity in tasks that require cognitive skills,” the BIS said. It cited a study by China’s financial technology giant Ant Group, which found its programmers were 55 per cent more productive when using a LLM to help with coding.
However, the BIS seemed less sure of what AI would mean for inflation, saying it could act as a deflationary force by raising workers’ productivity — as many have predicted — but also outlining a future in which it boosted prices by raising demand.
While pointing to the benefits for central banks of harnessing AI in their own operations, the Basel-based body also flagged several potential risks from the technology, such as incidences of it providing incorrect information and its vulnerability to hacking.
“AI will affect financial systems as well as productivity, consumption, investment and labour markets, which themselves have direct effects on price and financial stability,” it said.
“To address the new challenges, central banks need to upgrade their capabilities both as informed observers of the effects of technological advancements as well as users of the technology itself.”
Trained on large data sets, generative AI is capable of having humanlike conversations and producing unique content.
Many companies have seized on the technology to gain a competitive edge since its emergence last year, including in the financial sector. But central banks have been more cautious owing to concerns about reliability, legal risks and transparency.
The US Federal Reserve has started looking into how it could use artificial intelligence in its own operations, but officials there are proceeding cautiously and not considering its usage in any policy work at this stage.
The Bank of England said this year that it was using AI “to support and enhance” its capabilities, such as in trying to predict economic growth, banking sector distress and financial crises.
The Financial Times revealed recently that the European Central Bank had started using AI to speed up many of its more mundane activities, from drafting briefings and summarising banking data to writing software code and translating documents.
The BIS said there were limits to how much the technology could replace humans in central banks. “While it may be able to perform tasks that require moderate cognitive abilities and even develop ‘emergent’ capabilities, it is not yet able to perform tasks that require logical reasoning and judgment,” it said.
But the BIS identified several areas where central banks could benefit from AI, such as “nowcasting” systems to scan vast amounts of real-time data to spot the build-up of financial risks or to predict downturns.
Other uses include detecting money laundering. The BIS said its Project Aurora had tested AI’s ability to find dirty money in payments data and found “machine learning models outperform the traditional rule-based methods prevalent in most jurisdictions”.
However, it warned the technology also carried risks, such as when AI models were corrupted by “data poisoning attacks”, leaving them vulnerable to manipulation. Widespread use of AI could lead to bias and discrimination, raise data privacy issues and lead to a dependence on a few big providers of the models, the BIS added.
There could also be financial stability risks if a large number of financial institutions used the same algorithms. This “could amplify procyclicality and market volatility by exacerbating herding, liquidity hoarding, runs and fire sales”, it said.
