Editor's note:This is a guest article by Tim Mueller, Managing Director of Global Investigations and Compliance at Navigant.

Have some sympathy for the poor banker—really, please do.

In American society, few are more "deservedly" the target of public ridicule than the men and women who make their living managing other people's money. Some of the contempt may be justified, but in every bank around the world, there is a group of employees who truly deserve your empathy: those assigned to solve the problem of money laundering.

It is extremely difficult to know exactly how much illegal money is generated each year through drugs, terrorism, human trafficking, and other criminal activities. But at the high end of estimates, the numbers are staggering—roughly $2 trillion globally。

Investigators at the Financial Crimes Enforcement Network (FinCEN), a bureau of the U.S. Treasury Department, are theoretically responsible for blocking illegal cash flows. But in practice, the front line in the fight against anti-money laundering (AML) is staffed by employees inside banks.

As a congressman recently told me, the government has largely "outsourced" AML policing to banks. Unfortunately, these well-intentioned current enforcers have fewer resources at their disposal than those trying to cheat. That means they need all the help they can get, including from the business world's hottest buzzwords.

All haystack, no needle

Artificial intelligence and machine learning may still be flashy ornaments in some industries. But in the AML field, the question is not what they can do for banks and the financial industry a decade from now, but what they can do right now.

To understand this, one first needs to grasp why AML is so difficult.

Fraud is the most common pain point in the financial industry, but banks have long had the ability to handle it. Its problem is binary in nature—was that charge on the credit card legitimate—so the answer is much easier.

By contrast, AML is "all haystack, no needle."

Money launderers often have ample resources, using shell companies, shadow buyers, and other sophisticated methods available to the highest bidder to cover their tracks.

In the United States alone, AML spending inside banksamounts to roughly $23 billion annually. Yet over the past decade, FinCEN has fined these institutionsanother $23 billionfor failing to catch enough criminals in the act.

It is not that bankers do not want to find a solution, nor that they are unwilling to spend money. The problem is that the solution is simply too hard to find. Which brings us back to those buzzwords.

More than promises

Systems already exist that can automatically generate alerts—the first step in identifying whether a transaction is suspicious. The problem is that the vast majority of alerts are unproductive, meaning they are ultimately determined to be normal activity.

However, machine learning is showing great potential both to reduce these false alerts and to identify genuinely suspicious activity.In some cases, false positive rates as high as 90% have been reduced to 50%. This means investigators can spend less time manually reviewing alerts while improving the effectiveness of alerts, making them more likely to be submitted to FinCEN for formal investigation.

Ultimately, humans will still make the judgment, filing suspicious activity reports only when necessary. Machines are not replacing critical work; they are simply providing incremental gains in efficiency—which over time can both reduce AML spending and lower the fines banks pay for weak AML compliance, while giving experts more time to focus on catching criminals.

This added effectiveness has not escaped the notice of the most eager audience for help. Five U.S. federal agencies, including FinCEN, issueda joint statement last December on high-tech applications in the AML field, saying they "welcome these innovative approaches to protect the financial system from illegal activity." A month before that statement, the Monetary Authority of Singapore also gave itsofficial endorsement to the use of AI in AML。

When regulators—typically the most conservative members of the ecosystem—show such enthusiasm for this opportunity, the technology offers far more than empty promises.

Bankers as a group may not always be the most sympathetic characters. But those tasked with solving such a complex, massive, and costly global problem deserve good tools, and our understanding.