ICI Viewpoints

What’s the future of the financial services workforce and artificial intelligence (AI)? Adam Broun, president and COO of Kensho Technologies, believes that AI will help human beings become better at what they do. “We’re not replacing humans yet in any meaningful way; we’re making people more effective and efficient in what they’re doing,” Broun told the audience at ICI’s annual Operations and Technology Conference, held concurrently with the General Membership Meeting on May 22–24 in Washington, DC. “What we focus on are tasks that are more subtle, that underlie processes that help make humans more effective.”

Kicking off the session, Broun described how Kensho developed its “new economy” indexes and how AI comes into play. “It’s hard as an investor to invest in new, emerging industries with one click. What we’ve done is create algorithms that read transcripts, 10-Ks, and other filings, and make algorithmic determinations about the exposure of a company [to] the specific sector you’re interested in. These [indexes] are truly algorithmic. There is no human, no stock-picking committee,” said Broun.

Weighing the Regulatory Implications

This use of AI can raise unique regulatory questions—for example, how can firms communicate AI algorithmic processes to regulators? Broun warned that it’s not a simple task. “It’s hard to have an algorithm ‘explain’ itself,” said Broun. Understanding what an AI program has learned and why a decision has been made is extremely difficult—as hard to explain as explaining how a human brain makes a decision, he said.

But, Broun pointed out, there is a lot AI can do that doesn’t have financial regulatory implications. AI can “read vast amounts of text data—news, filings, transcripts, call notes—and allow an analyst to ‘smart search’.…It then learns what I’m interested in and suggests new pieces and documents and segments within the documents. It’s not making decisions, but making that analyst massively more productive, and able to churn through millions of pages in a matter of minutes.”

Unlocking the Value of AI to Operations

This sort of application is where AI currently can be most valuable to financial services, explained Broun. “If you stand back and look at a task somebody does that’s stupidly hard, takes a stupidly large amount of time to do something simple…then that is an area where the technology can help you,” he said. Focusing on uses in the operations arena, Broun outlined where AI can be extremely helpful: mass data management, compliance, risk management, onboarding, know-your-customer requirements, and anti–money laundering procedures. “Anywhere there are repetitive tasks that humans are doing because there is some noise or dirt in the data,” such as in customer data, “AI can outperform humans,” he said.

Broun shared an example of where Kensho Technologies trained AI to clean up a new data set. The old way of cleaning up this data involved doing it manually, by sending it offshore and having 100 people comb through it over many months. With the company’s AI solution, “we did it in one weekend with two people. It is thousands of times faster and cheaper,” said Broun.

“The pitfalls,” warned Broun, “are the cost of not doing this.”

What was the biggest thing that Broun wanted the audience to understand about the potential effect of AI on the workforce? He said that people tend to overestimate the short-term impact of AI and underestimate its long-term impact. “Short-term, it is going to be around the edges, on subtle processes that make us more productive. What humans tend to do is adapt, and they shift work around and find new things to do because they have more time to spend. And I’m not going to bet against that,” Broun said.