SEC Division of Economic and Risk Analysis Acting Director and Chief Economist Scott Baugess discussed the development and use of artificial intelligence and machine learning within the SEC, challenges presented by artificial intelligence technologies, and the future of these technologies within the SEC.
Mr. Baugess explained that progress in development of artificial intelligence technologies has made it necessary for regulators to examine the potential uses and impacts of this technology on the regulatory environment. He observed that while there is obvious value in potentially being able to more effectively predict investor behavior, “latent variables,” such as fraud which is not seen until it is found, make understanding likely outcomes an especially difficult task. Because of these unobservable outcomes and other difficulties such as translating languages, the application of machine learning to regulating financial markets is less straightforward than it is in other contexts.
Machine learning has been utilized by the SEC in various capacities, including to analyze tips, complaints and referrals and to identify abnormal disclosures. Mr. Baugess noted that machine learning is also useful for detecting potential investment adviser misconduct by identifying outlier reporting behaviors. While acknowledging the value of this form of analysis, he cautioned that reliance on machine learning technologies, such as feeding the results of unsupervised learning algorithms into machine learning, can lead to false positives, or instances where misconduct or SEC rule violation is errantly identified.
Mr. Baugess concluded by emphasizing that although machine learning will improve the SEC’s ability to identify possible fraud or misconduct, he expects that “human expertise and evaluations” will always be necessary in the regulation of capital markets.
Lofchie Comment: This is technology expertise that the SEC should consider outsourcing. Perhaps the SEC should call up the credit card companies and ask them for some lessons in catching fraudulent transactions.