House Financial Services Committee Examines AI Implications for Financial Services Industry
Commentary by Steven Lofchie
At a hearing on "The Impact of Artificial Intelligence ("AI") on Capital Markets and Jobs in the Financial Services Industry," the House Financial Services Committee considered expert testimony on AI's impact on investing and compliance and risk management.
Automation of Investment Decisions
Vice Provost of NYU and Professor of Media, Culture, and Communication Charlton McIlwain warned that "regtech" (i.e., a combination of automation and algorithmically driven systems) is susceptible to (i) reliance on inaccurate or unrepresentative datasets and (ii) humans being unable to audit the processes used in algorithms to process data that potentially could lead to "disparate outcomes."
Dr. McIlwain asserted that biases in algorithms against people of color are not solely a technical problem in AI automation, but a racially discriminatory problem as well.
Modern Markets Initiative CEO Kirsten Wegner expressed support for automated trading, saying that it had lowered costs for investors to a "fraction of the price," as compared to a decade ago.
Nasdaq Market Surveillance Head Martina Rejsjo recommended that financial firms consider whether adopting AI and big data for investment processes is worth the substantial costs. Additionally, she advised firms to compare AI performance against traditional techniques.
The Financial Services Workforce
Dr. McIlwain advocated for the development of computer-driven automation that protects labor and civil rights.
Cornell University Professor of Practice at the School of Engineering Marcos Lopez de Prado emphasized the importance of teaching machine learning ("ML") and automation to workers in the financial sector.
Ms. Wegner advocated for assurances that AI and automation should be a resource and not a replacement for humans.
Chartered Financial Analyst Institute Senior Director Rebecca Fender stated that while 89% of industry leaders believe that workers' roles will "be transformed multiple times during their careers," AI will not replace all skills. She noted that AI has still not been able to perform skills such as "ethical orientation, transparency, communication, empathy, tacit knowledge, and trust interaction."
Compliance and Risk Management
Dr. Lopez de Prado advised regulators to use ML to detect market manipulators and false investment products. He also defended ML algorithms, saying that while they may incorporate human biases, it is easier to monitor and modify them than "relying entirely on humans."
Ms. Wegner urged financial regulators to gather the "funding resources, technological capacity, and access to AI" in order to compete with the growing sophistication of "bad actors" in the financial markets.
Ms. Rejsjo emphasized the importance of human input in both analyzing the surveillance systems' output and training surveillance machines to increasingly put out more accurate information.
Commentary
Arguments about whether firms "should" use AI are largely futile and irrelevant.* It's a bit like arguing over whether to use that four-wheel contraption, when a good horse will get you there in a few days. Likewise, demands that AI be a "resource" and not a "replacement" for humans are just words that sound caring, but that have no real substance.
AI will be used where it can do a better, faster, more efficient job than humans can do. Should U.S. legislators choose to create economic impediments, others may not, and those that do not will simply out-compete the United States. Assuming that AI will be used where it is economic to do so, however, does not end the questions. It's actually a starting point toward more meaningful questions about the future. Debating about whether firms should use AI does not help.