This month our RegTech 2021 series, continued by examining government uses of artificial intelligence (AI). Just last year, Congress passed legislation encouraging the government to move from pilots and demonstration projects to scaled up, applied solutions. The discussion featured two fireside chats with government leaders: Henry Kautz, Division Director, Information & Intelligent Systems, National Science Foundation (NSF) and Mike Willis, Associate Director in the Division of Economic and Risk Analysis, Securities and Exchange Commission (SEC).
First Director Kautz discussed the work in AI at NSF, as the agency seeks to fund larger, more interdisciplinary projects. Lately, the agency has been focused on establishing AI Institutes, virtual centers organized around themes, connecting colleges and universities to partners in the private and public sector. Themes include AI-Augmented Learning, AI-Driven Innovation in Agriculture and the Food System, and Advanced Cyberinfrastructure. Director Kautz emphasized the importance of NSF’s role in supporting foundational, pre-competitive research and development in these private-public partnerships.
When thinking about what challenges the government is facing, he recommends that agencies consider improving coordination among themselves on how to best make use of AI internally. He pointed out the success of coordinating bodies like the Joint Artificial Intelligence Center at the Department of Defense, but encourages the government to think more broadly about the big questions facing the government. Additional suggestions to scale up AI include building up AI expertise within the government, especially at the managerial level, being sensitive to and aware of AI skepticism, and rethinking traditional procurement practices. He also emphasized the need for explainability and transparency in ensuring ethical uses of AI and conceptualizing data as infrastructure.
In the next fireside chat, Preethy Prakash, Vice President of Business Development at eBrevia, spoke with Mike Willis from the SEC. Willis, speaking for himself, spoke of the SEC’s steps to make registrant disclosures more accessible and usable, after noticing well over 90% of EDGAR visitors are machines.
Even though these data sets are highly desired by the public and outside uses of AI, the role of AI within the SEC today is largely focused on the enhancement of the effectiveness of staff analytical procedures, including those related to risk assessments, identifying potential areas for further investigations for activities like insider trading, comment letter analysis, and entity mappings.
When asked how to think about creating quality data that is interoperable, Willis pointed directly to the Evidence Act which defines the term “open government data asset” based upon an underlying open standard. “Leveraging industry and market standards, I think, are a very useful way to drive down compliance costs, while streamlining the validation and analysis of the data, including for AI and ML purposes,” Willis stated. He went on to note how these open standards are a great example of public-private partnerships discussed previously.
As the SEC continues to implement AI, Willis outlined some change management considerations. His recommendations were to ensure that you have talented, qualified professionals, help people understand the problems and processes that AI can help supplement, provide use cases and examples, ensure that your AI solution stays within its scope, and finally, echoes Director Krautz’s call to consider data as infrastructure, meaning it be standardized and structured.