Unless you’re a sharecropper or a hermit, or you live off yak meat in the wilds of Alaska, you are connected. And if you’re connected, you might think you’re using technology, but in fact it’s using you. And guiding you. Leading you by the nose, in fact.
How often have you heard these things from your digital tools?
- I think this is spam.
- These results are the most likely ones to matter to you.
- If you liked that, you might want to buy this. Other people as ugly as you also bought this other thing.
- If you’re heading to Cleveland, you probably want to use this airline.
- Your ride will probably cost this much, and it will pick you up in the next seven minutes. If you drive yourself, it will take you this long if you take this route, less time if you take this other route. Wow, you drive like an idiot, so if you don’t die, you’ll get there ten minutes earlier.
- You wanna tag that face? She looks like your sister.
- You may know these other people. Do you want to connect?
These are all examples of machine learning. Robotic processes assimilate behavior, preferences, queries, hat size, etc., and then make recommendations. These are cute, and they might be even be useful at a tactical level. They can provide context to internet fashion influencers, who will be the first people eaten when the apocalypse comes.
In the comics, Jimmy Olsen had a special watch that could call his friend Superman in an emergency. But he would use it when he was in danger, not when he couldn’t open a jar of pickles. The little things are handy, but the bigger things are more impactful. Archimedes famously said, “Give me a lever and a place to stand, and I will move the world.” He didn’t say, “I will move the couch so I can vacuum under it.”
So instead of recommending the best battery charger or hand cream or shade of blush, what if machine learning tells you which of your thousands of potential customers you can sign up for a credit card or loan, and which ones will steal from you? Doesn’t that seem more useful?
It’s one thing to size up known good and bad applicants so you can say, “I’ve seen that person before, and I know how risky or non-risky they are.” Definitely there is value in that. That’s how credit bureaus do their thing. But what if:
- That person isn’t really that person, but rather an identity thief?
- That person keeps changing their Personally Identifiable Information to get different results after multiple rejections?
- That person isn’t really a person, but a synthetic identity designed to max out credit cards?
- That person is brand new to the working world and isn’t in your repository?
So now your ID database isn’t going to do the job. That’s when machine learning is your friend. By ingesting all those goods and bads, I learn the patterns. I don’t necessarily say, “I’ve seen that guy before.” I say, “I’ve seen people like that guy before. I can model my response based on that pattern or model.” Your Tesla may not have driven the road you’re on, but it knows what a stop sign, a traffic signal and a double yellow line are and can react appropriately even on a strange road.
This is the value of machine learning to fraud prevention and identity verification. At Socure, our machine learning models take in literally millions of applicants and accompanying identities on a steady basis. And because we use a larger volume and variety of data than anybody – more than any other partner, customer, vendor, telco, anybody – our models learn more, and therefore produce better predictability than any other solution on the market. It’s all about:
- Auto-accepting good applicants, to get revenue flowing quickly
- Deflecting bad applicants, to avoid the costs of fraud
- Automatically shrinking false positives and the application of friction because of those first two points
Our models auto-approve 95%. We capture far more fraud, and we drastically reduce the need for friction. More revenue for the top line, less fraud and friction for the bottom line. We also provide KYC. And it’s all because our AI-driven platform learns, and keeps learning, how to learn even more.
Figuring out which hand cream to buy? You’ll have to learn that on your own. Don’t sweat the small stuff. And let us sweat your big stuff.
Jeff Scheidel is a technologist with 38 years in software, including 26 years in security solution design. He is the author of numerous white papers on security and regulatory compliance, as well as a McGraw-Hill book on identity, access, database, and application protection. Jeff is an expert on compliance requirements across a number of industries, and has presented at a wide variety of security events.