Treasury official discusses AI and cloud computing at Gov2Gov summit
On October 24, Assistant Secretary for Financial Institutions at the U.S. Department of Treasury Graham Steele delivered remarks at the Gov2Gov Summit to discuss the benefits and risks of artificial intelligence (AI) and machine learning (ML) in the financial services sector.
First, Assistant Secretary Steele discussed the role of cloud computing and cloud service providers (CSPs) in supporting financial institutions’ work, following the Department’s release of a February report which discussed the financial sector’s adoption of cloud services. Assistant Secretary Steele indicated, among other things, that while cloud services can offer more scalable and flexible solutions for financial services institutions to store and manage their data, financial institutions have struggled to understand clearly and implement the cloud services they are purchasing from large, market-dominating CSPs. Assistant Secretary Steele stated that the Department is working toward a model that will allow financial institutions to “unbundle” cloud service packages so that financial institutions can provide more individualized services.
Next, Assistant Secretary Steele discussed the potential advantages and disadvantages of the use of AI among financial institutions, which use AI for tasks including credit underwriting, fraud prevention, and document review. Among the benefits AI offers to financial institutions are reduced costs, improved performance, and the identification of complex relationships. The risks of AI, according to Assistant Secretary Steele, fall into three categories: (i) the design of AI, which can raise discrimination concerns, such as in consumer lending; (ii) how humans implement AI, including the possible overreliance on AI to render financial decisions; and (iii) operational and cyber risks, including the dangers around data quality and security, as AI consumes significant volumes of data.
Last, Assistant Secretary Steele discussed how policymakers are addressing privacy and discrimination concerns with AI. He mentioned the White House’s Blueprint for an AI Bill of Rights, which would require, among other things, regular assessment of algorithms for certain disparities and biases. Assistant Secretary Steele also cited regulatory actions that can address the risks of AI, including a CFPB rulemaking under the FCRA and Federal banking agency guidance on third party risk management.