They demonstrate that AI can be an ATM for central banks.

Source: Criptonoticias Original Title: They demonstrate that AI can be the ATM of central banks Original Link:

• ChatGPT decided payments and liquidity in seconds without prior training for central banks.
• AI can do in banks what DeFi has been doing for a while in the world of cryptocurrencies.

An innovative study by the Bank for International Settlements (BIS) demonstrates that generative artificial intelligence (AI) agents can perform critical liquidity management functions in central banks and high-value payment systems, traditionally handled by humans.

The research, conducted with the reasoning model o1 of ChatGPT in agent mode, simulated real scenarios where the AI had to balance liquidity costs and risks of delays in multimillion-dollar transactions.

The experiment designed three scenarios that replicate real challenges in RTGS or real-time settlement systems (Fedwire, TARGET2, Lynx, etc.), the heart of the traditional financial system.

In the first scenario, the AI had only 10 dollars in liquidity and two pending payments of 1 dollar each. Faced with the possibility of an urgent order of 10 dollars, it decided to freeze everything. Its own explanation made it clear why it made the decision: “I delay the small payments now to preserve liquidity and be able to handle the urgent transaction if it comes.”

The second scenario introduced greater complexity with probabilities of receiving external funds (90%) and executing urgent payments (50%). In this case, the AI only processed lower-risk transactions, demonstrating dynamic prioritization capability.

The tests showed that even when varying probabilities from 50% to 0.1% or scaling amounts up to billions of dollars, the AI maintained its cautious approach. However, in complex situations, its consistency slightly decreased, with occasional variations in decisions.

AI is already a better treasurer than most humans, says the BIS

The study proposes to develop “AI assistants” for routine tasks, reserving human roles for supervision and strategic decisions. Researchers project that similar systems could be tested in regulatory sandbox environments before real implementations.

“The results suggest that specific AI solutions could reduce operational costs and improve efficiency and operational security,” the BIS report indicates. But it warns of limitations: the models depend on historical data and may fail in the face of extreme events or “black swans” outside their trained experience.

The study compares this approach with traditional reinforcement learning. The authors highlight that, unlike traditional reinforcement learning ( which requires thousands of simulations), generative AI achieved “excellent results with zero specific training.”

So for that level of effectiveness, the authors of the report believe that AI could save millions in immobilized liquidity and dramatically reduce payment queues in RTGS systems.

Although the BIS report focuses on traditional financial systems, its findings are not surprising in the world of digital assets. This is because decentralized finance applications (DeFi) have been managing liquidity in a 100% automated way for years with automated market maker pools (AMM), flash loans, and algorithms that rebalance in seconds.

What the BIS celebrates as innovation, certain DEXs, DeFi lending protocols, and other platforms have already been doing since 2020 with billions of dollars at stake.