Regulators want to create a fair and orderly marketplace. The problem is that the idea of “fair” is fungible and in a capitalist system, “orderly” can be nebulous.

Still, regulators are trying to ensure that a level playing field exists, though that definition changes with each jurisdiction and regulatory body. The problem is that as new technologies work their way into the financial services arena, regulators are struggling to keep up, particularly when it comes to the use of artificial intelligence (AI). 

While regulators are well-versed in market structure challenges, do they understand AI well enough to implement more prescriptive guidance—or even rules—in this constantly evolving field of technology? Regulators will always be behind the eight ball when it comes to AI (and, more specifically, machine learning), and must be cautious in their approach to how firms use AI so as to avoid stifling innovation by mistake.

One recent example that led to some banks fearing that regulators could take a more rigid stance on how AI is used within capital markets is the US prudential regulators’ request for information (RFI) on the uses of AI and machine learning. 

Led by the Federal Reserve, US banking regulators are trying to understand how financial firms use AI, in which areas of their business, and the challenges they experience in developing, adopting, and managing AI. They also want to know if clarification from the agencies would be helpful for these institutions to use AI in a “safe and sound” manner.

The deadline for comment closed on July 1, and according to the agencies, submissions will later become part of the public record and subject to public disclosure. It should be noted that for the time being, when we talk about finance, we’re not talking about the institutional/wholesale capital markets.

Other regulators are trying to be more proactive in understanding AI. For instance, there’s the Monetary Authority of Singapore’s (MAS’s) Veritas initiative, which will enable financial institutions to evaluate their AI and data analytics solutions against the principles of fairness, ethics, accountability and transparency (Feat). 

The central bank recently launched the Global Veritas Challenge, a competition seeking to accelerate the development of solutions that meet the Feat principles. Fintech firms, solutions providers and financial institutions globally have been invited to submit innovative solutions to address eight problem statements identified by banks. These focus on validating the fairness of AI solutions for banking use cases in product marketing; risk, compliance and fraud monitoring; loan origination and know-your-customer (KYC); and credit scoring and profiling.

Applications for the challenge closes at the end of July. But, again, we’re not talking (yet) about hardcore trading platforms and models.

I recently interviewed Shameek Kundu, former chief data officer at Standard Chartered, on a recent episode of the Waters Wavelength Podcast. We spoke about various topics, including regulators’ approaches to AI and machine learning. Kundu says he believes that, so far, those approaches have been thoughtful and nuanced.

“I genuinely think every regulator that I’ve spoken to—and probably across the world, there’s at least eight or nine major jurisdictions that I’ve spoken to on this topic—is approaching this in an extremely thoughtful and nuanced manner,” he said.

My chat with Kundu reminded me of a previous podcast episode when I interviewed David Hardoon, formerly the chief data officer and head of data analytics group at the MAS before joining UnionBank of the Philippines, where he’s now the senior advisor for data and AI. He told me there is a balance between innovation and governance. 

Hardoon said one of the greatest things he learned during his time at the MAS is that “there is truly a harmony and a balance of innovation and governance.” 

“I’ve learned and appreciated that in fact, good governance, [and] good regulation results in phenomenal innovation. It basically is a breeding ground for the possibilities. It’s not ‘What can you do?’ but asking the question of, ‘What should you do?’” Hardoon said.

There’s rapid digital transformation in the industry, and regulators—like financial firms—need to understand these tectonic shifts. 

Skeptics, good and bad

Humans are curious beings by nature. We ask questions, we seek information, and we are skeptical—for better or worse. The Covid-19 vaccine rollout is a prime example of this: It’s a miracle of science to have a vaccine so readily available to stem the spread of Covid-19, but because humans are skeptical and seek out information—and sometimes that information (or, in this case, disinformation) is very bad—they don’t get a vaccine, despite all the scientific evidence that it will protect them.

This brings us back full circle, I think, to what Kundu believed is the barrier to the adoption of AI within the capital markets: the lack of trustworthiness. 

“There’s a whole question of, ‘Do we understand all the machine learning models?’” he asked. 

It’s a learning process for trading firms and regulators, alike. But, there will always be bad actors in the market. 

Should regulators dig deeper on the potential threats of AI, delving further than the typical biases and ethical questions? Right now, there are more questions than answers. But since this is a rapidly developing space, I want to believe (and I hope) that the use of AI produces more good than bad. Hence, regulators should be cautious about being more prescriptive so they don’t hamper innovation. 

Or, as Kundu told me on the podcast: “I do think after two or three years of thinking about it, perhaps some of them will perhaps become more prescriptive in their guidance. But from every account I’ve had so far, it should not be something that stifles innovation too much. Of course, it will increase a level of governance and discipline as time goes by, but that’s to be desired.”

I hope that turns out to be the case.