Jeopardy! viewers in 2011 witnessed a first in game show history: a computer not only competed in the show, but wiped the floor with its opponents. Watson, a cognitive computer system, left Ken Jennings and Brad Rutter—previously Jeopardy!’s top champions—in its dust over a three-round matchup.
IBM built Watson specifically to compete in a trivia show, but the potential applications proved highly commercial. Watson is programmed to understand human language questions; comb through massive amounts of data; and provide intelligible, quick and correct answers.
Unlike humans, Watson remembers everything. With every question and answer, Watson gets a little bit smarter. Through machine learning—by which computers gain capabilities without being explicitly programmed for them—Watson can become a technical expert in a variety of fields. In 2013, for example, Watson was deployed for lung cancer treatment decisions at Memorial Sloan-Kettering Cancer Center in New York City, incorporating symptoms and crunching data to recommend treatment approaches with specified levels of confidence.
Crucially, Watson did not supplant doctors’ decision-making. Instead, Watson became like a very knowledgeable colleague in the exam room, providing insights and making connections the health care providers might not have recognized otherwise.
Last year, IBM acquired Promontory Financial Group, a D.C.-based consulting firm led by former regulators and industry veterans, to bring Watson’s expertise to bear on financial regulatory compliance challenges. Promontory founder Gene Ludwig, a former comptroller of the currency, was excited by the opportunity to bring Promontory’s in-house expertise to banks of all sizes in addition to the larger institutions that often hire the firm.
“For community and regional banks, the promise is that over time they will have access to our thoughts in an efficient and effective manner, and for the largest institutions, they have a new colleague called Watson,” he says. Promontory is in the process of teaching Watson about regulations and guidance, increasing Watson’s ability to understand the compliance questions bankers will ask it.
Banks using Watson can use its capability to help their compliance teams connect regulatory requirements to obligations to controls, then manage those controls through an interactive dashboard. For example, a compliance officer might review a document for relevant obligations. Instead of remembering or manually looking up each relevant regulation, Watson can scan the document and highlight specific areas where regulation or guidance is likely to apply. The banker can then review those recommendations and, much more quickly than before, incorporate them into the compliance program.
“It’s augmented intelligence, not artificial intelligence,” says Alistair Rennie, general manager for Watson Financial Services. “The more we can look at in context, the better we can get answers to be, and the more we can work through expert guided discussions about what the right answer, decision or choice is, the better everything gets.”
Today, compliance management systems are often based—conceptually—on a stack, with every new regulation layering on yet another obligation or control onto a product in order to ensure that no obligation goes unmet. Layering is an inefficient way of managing compliance risk, because in seeking comprehensiveness, it introduces duplication. A cognitive computing system can instead envision networks of intersecting requirements, ensuring the same comprehensiveness but focusing attention on the network nodes that most warrant it.
The sheer computing power of systems like Watson enables banks to bring enormous volumes of data, both structured and unstructured, into their compliance programs. For example, Watson can contribute to anti-money laundering and “know your customer” compliance by incorporating social media data, says Grace Brasington, an IBM VP serving as global leader for risk and compliance for Watson Financial Services. It can provide “voice review on every call” for insider risk management and internal reviews, and it can tie in transaction data. It can replace much of the time-intensive manual review, freeing up compliance officers’ time for higher-level reviews and strategic priorities.
The future of the compliance officer
So—are compliance officers about to be replaced by systems like Watson? That’s how Ken Jennings felt when he lost to Watson on Jeopardy! “I felt like a Detroit factory worker in the ’80s who could see a robot do his job on the assembly line,” he reflected. “Quiz show contestant was the first knowledge job that had become obsolete under this new way of thinking.”
But while Watson may have been designed to win in the zero-sum world of Jeopardy!, it wasn’t designed to supplant human capacities in banks. “Nothing we’re doing on cognitive capability displaces current systems,” notes Brasington. “It makes them more useful.”
Rennie notes that IBM has designed the system around Watson to be evidence-based. Watson doesn’t just spit out an answer; it “lays out the evidence in a way that an expert can understand. It’s not a black box.” That still requires human compliance officers doing strategic thinking and deep analysis.
Ludwig thinks cognitive computing can make compliance jobs “a great deal more satisfying” as “the tool is constantly refined to ensure the compliance officer is able to do her job much more efficiently and effectively than would be possible without the technology.”
By taking freeing up time and brainpower for compliance officers to think strategically, Rennie believes that technology like Watson can help “move us to a world where compliance is viewed as a platform for innovation.”
This article originally appeared in the May/June 2017 issue of the ABA Banking Journal and is reproduced here by permission.