Jon Denilson
Jon Danielsson
Director of the Centre for Systemic Risk, London School of Economics and Political Science
Andreas Utman
Andreas Uthemann
Principal Investigator at the Bank of Canada and Fellow at the Centre for Systemic Risk at the London School of Economics and Political Science
If financial regulators are proactive in engaging with artificial intelligence (AI), it is likely to be of great help to these institutions. But if these institutions are conservative and reluctant to participate, they risk being rendered useless and at risk of financial instability.
The private sector is rapidly adopting AI, although many financial institutions say they intend to err on the side of caution. Many financial institutions have large AI teams and have made significant investments. JP Morgan reports spending more than $1 billion a year on AI, while Thomson Reuters has $8 billion in AI funding. AI helps financial institutions make investments and perform back-office tasks such as risk management, compliance, fraud detection, anti-money laundering, and "know your customer." AI is expected to deliver significant cost savings and efficiency gains, and its application will inevitably grow rapidly in a highly competitive financial system.
As the private sector adopts AI, it will also be more responsive and help find loopholes in regulations. As we point out in our book by Danielsson and Uthemann, regulators must keep up with the times if they are to continue to function.
Regulators have so far been slow to adopt AI approaches, and they will find it challenging to adopt AI. This will require adjustments in institutional culture and personnel, a change in the way they are regulated, and a significant allocation of resources.
Give full play to the advantages of artificial intelligence
In many areas, AI can help financial regulators function.
AI can help small regulators design rules and regulations and enforce compliance with them. While human supervisors will initially make executive decisions, reinforcement learning through human feedback will help AI supervisors become more and more efficient to perform autonomously. Adversarial architectures, such as generative adversarial networks, may be particularly useful in understanding complex areas of official and private sector interactions, such as fraud detection.
AI will also help large regulators, for example by making recommendations on how best to respond to stress and crises. AI can simulate options for coping with stress, recommend and implement interventions, and analyze the drivers of extreme stress. Regulators can use generative models as artificial laboratories to experiment with policies and evaluate private sector algorithms.
AI will also be useful for general economic analysis and forecasting, leveraging public and private data, established economic theories, and past policy analysis to augment common underlying models through transfer learning. Reinforcement learning with feedback from human experts can help improve the engine. This kind of AI can be useful for economic forecasting, policy analysis, and macroprudential stress testing.
AI will be of great help to financial institutions, but they also face a greater risk of losing control as a result of AI.
If financial regulators are to continue to effectively supervise the financial system, they must change the way they operate. Many institutions will find this challenging. AI requires new ways of regulation, requiring different approaches, human capital, and technology. The high cost of AI and the oligopolistic nature of AI suppliers pose particular challenges. If regulators are unwilling and slow to engage in AI, they run the risk of being wiped out.
However, when regulators adopt AI, AI should be of great benefit to their functioning. Microprudential regulations are designed and enforced to benefit from this because the abundance of data, relatively immutable rules, and clear goals all help AI to its strengths.
At the macro level, it's even more challenging. AI will help scan the vulnerability of the system, assess the best measures to deal with stress, and find the best crisis interventions. However, it also brings with it the threat of AI illusions, and therefore inappropriate policy responses. Measuring the accuracy of AI recommendations is critical. It would be beneficial for regulators if they were willing to adopt a consistent quantitative framework for measuring and reporting the statistical accuracy of their data-based inputs and outputs.
Regulators need to recognize the benefits and threats of AI and incorporate this awareness into their practices in the performance of their social service functions.
Original link:
https://cepr.org/voxeu/columns/how-financial-authorities-can-take-advantage-artificial-intelligence
(Chang Changsheng/excerpt)
The above views and remarks do not represent the position of this platform.
===中国智库 China Think Tank===
Thank you for your attention and dissemination of "China Think Tank"
We strive to deliver valuable returns
Chinese think tank
China Think Tank is a serial publication under the academic guidance of the Development Research Center of the State Council and edited and published by China Development Press. It aims to let policymakers know more about the voice of the people, make the content and process of decision-making more recognized by the public, and build a bridge between the government, the theoretical community and the people.
"China Think Tank" is a new media account set up by China Development Press. China Development Press is a national high-end think tank directly under the Development Research Center of the State Council, and its business areas include book publishing, periodical publishing, research consulting, conference exchanges, think tank parks, etc., and is committed to building a first-class think tank-type publishing and media group. In addition to the publishing house, it also has periodicals such as China Development Observation, China Economic Yearbook, China Economic Report, and New Economy Tribune, as well as Guoyan Culture Media Group Co., Ltd. (Guoyan Think Tank), Guoyan Think Tank Innovation Science Park Investment Co., Ltd., Beijing Guoyan Think Tank Management Consulting Co., Ltd., Beijing Zhiyan Future Information Consulting Center, Shandong Institute of High-quality Development and other institutions.
Think tank cooperation: Mr. Xu 18101277387
New media cooperation: Ms. Du 010-67889142 18813067381