When we think of robots, fictional characters like R2-D2, Optimus Prime, Hal, or Bender might come to mind. But, today’s robots aren’t cute sidekicks, superheroes, or destructive forces—and they’re certainly not fictional.
Last month, Socure’s founder and chief strategy officer, Sunil Madhu, talked with PYMNTS.com’s CEO Karen Webster about how the current process for identity verification is broken. Madhu discussed how machine learning can significantly advance identification verification and how Socure is already making great strides in this regard.
The entire hour-long PYMNTS webinar, titled “Fixing Digital Identity Verification: Robots to the Rescue,” can be viewed here and is definitely worth a listen. Here are a few highlights:
- The current (vastly outdated) process for ID verification relies on credit bureaus and essentially requires that we have to get ourselves into debt in order to prove ourselves to be trustworthy. That shouldn’t be the system we rely on—and it doesn’t work any more. Millennials, one of the major current demographic groups, use credit far less than previous generations, and the 1.5 billion of them who have “thin files” run into problems when it comes to credit bureau verification. Moreover, 180 countries around the world don’t have credit systems, which poses a huge problem for immigrants in the United States trying to open accounts with banks and other institutions. What’s more, the information credit bureaus use—like names, dates of birth, and Social Security numbers—are no longer private. They can easily be accessed on the black market—over 2 billion identities were stolen in 2017— allowing fraud to run rampant. As a result, credit bureau verification ends up rejecting many legitimate applicants. “If you’re going to rely on a system that rejects 50% of people knocking on your door, there’s a problem there,” Madhu says.
- The future of identity verification must be dynamic, frictionless, and free of bias. People’s identities change over time, and so should the data that verifies them. Moreover, people should be free to move around the internet without having to remember their high school mascot or best friend from first grade.
- Machines will always do a better job than humans, for a number of reasons. For one, humans are inherently biased—but machines don’t have to be. “If you take a racist human and you try to train racism out of them, you’ll find that it doesn’t work,” Madhu says. “A lot of facial recognition systems have trouble differentiating between people who are not white because the training for the machines doesn’t use nonwhite people. But if you supply the system with other faces, the system will learn. You can’t do that with human beings.” Humans are also inefficient and slow, while machines can process vast amounts of data very quickly. Even rules-based systems rely on human intuition and are not ideal; rules are reactive and predictable and can be easily circumvented.
- Socure’s ID+ solution provides a solution to all of these problems. In the works for several years—and currently being used at top banks and other companies. By drawing on a wealth of online and offline data, and piecing together shards of information, the solution can easily verify identities, eliminate bias, and reduce fraud. “It’s an export system, it’s not conscious, but it understands a lot about what it means to be a real person versus a risky identity online and in the real world,” Madhu says.
- Fraud, acceptance, and manual review are three sides of the same identity verification triangle. “Typically, humans can optimize for two sides,” Madhu says. ID+, however, can optimize for all three.
- AI perform better than humans because they reduce bias, have much higher accuracy, and work much faster. As a result, they can help solve the identity verification crisis by increasing business growth, cutting down on fraud, and streamlining the review process. AI solutions have been proven to reduce fraud by 60 to 85%. With ID+, Socure has been able to provide the lowest fraud rates in the industry, transforming companies’ processes—and results—almost instantaneously.