The Role of AI in the Identity Verification Process

Is AI Worth The Hype? 

Yes. If anything, the importance and potential of Artificial Intelligence in the KYC space is underplayed and underrated. The industry has been under immense pressure in the past couple of years with regulations and scandals redefining the means and ways you go about Know Your Customer procedures. 

In the midst of this disarray, AI has emerged as the guiding light, the savior the compliance space was calling for. What does AI mean for compliance, what’s the role of this new3 technology and what problems does it really solve? Let’s investigate. 

Reassignment & collaboration

The idea that Artificial Intelligence will replace humans is ignorant, to say the least. Can AI machines sort through the tons of data better than humans? Yes. Does that mean machines will simply replace the people that have manually performed these tasks up until now? No. It’s important to understand that AI technology is not something you plug in and identity verification problems simply go away. 

AI technology is an ecosystem, it’s a framework built around people, their experience, skills, and ability to make sense of the data analysis and findings. People in the KYC, AML, and identity verification industries would not get replaced, but reassigned. Their roles would simply develop and evolve around the new technology. AI is not something that would take their jobs but a powerful tool that would help them do their job better. 

Identity verification is a problem that tantalizes companies across the globe. It’s a universal problem that has been dealt with in isolation up until now. One of the main ideas proposed by AI technology is the ability to deal with this problem on an international basis. Technological consortia may just be the solution to the problem. A scenario where banks and companies share their data, creating a centralized information center that is way more powerful than what each company could have ever achieved on its own. 

An appropriate term for it could probably be collective intelligence. A collaboration that benefits all partakers without meddling with competition principles or profit generation. 

AI & UX Are Best Friends 

Want some double-barreled words to describe the current KYC and identity verification process? Labor-intensive, time-consuming and error-prone. Unfortunately, all three words perfectly describe the process most financial institutions and companies use to onboard their clients. 

What’s interesting is that we always look at the onboarding process from the business side of the equation. Rarely do we take a step back to evaluate the user experience and how customers feel when they go through the long and arduous KYC process. 

In a world where everything is becoming mobile-first and people are used to interacting with apps, clicks, taps, and touches, the identity verification process for a customer that wants to open an account with an FI is long, manual and arduous. Many companies are subject to worrying drop off rates as users can’t be bothered to go through the entire onboarding process. In the UK, 25% of applications are abandoned due to KYC friction according to a 2017 report from Consult Hyperion. 

This is where AI can work its magic and not only simplify the process but make it modern, fresh and minimal. Manual checks such as photo/ID verification can happen within minutes and people can be onboarded on the system within hours. Running a name through huge databases, PEP lists, and adverse media articles used to be a long and exhausting task for compliance staff but it’s a basic task for AI. The automation of this process significantly minimizes the time people have to wait for being onboarded. 

Better user experience does not only create healthier customer retention numbers for companies but it builds a relationship of trust between user and company. A fast, hassle-free onboarding process leaves a sense of satisfaction and content in the customer. It builds a foundation for a bright future as the first time the customer was asked to interact with the company they walked away with a positive impression. 

Consolidate Data Silos

AI is known for going through humongous numbers of data and information, organize and structure them in a short period of time. Even though that’s impressive, that’s not where it stops. Through Natural Language Processing (NLP), it is possible to not only compile data but make sense of it. 

AI has the ability to comb through data and build risk profiles with just the relevant information about an entity, cutting through the noise and anything irrelevant. Apart from the obvious time-saving implications afforded by AI in this scenario, companies will also end up with more robust risk profiles - faster execution and better results all in one. Analysts will now be presented with an up-to-date, detailed yet relevant distillation of what they need to know about an entity without having to do any manual work. 

Mitigate False Positives 

Monitoring the transactions of a suspicious entity is a tricky affair. Manually following these entities is logistically impossible so FIs resort to a rules-based system where they set certain parameters that the entity needs to operate within. If for example, the system recognizes a transaction that is above the set parameter, it will flag the transaction and alert the FI. 

The problem with this system is that the rules-based approach is too simplistic and does not take into consideration variables that might explain these transactions. Even though a transaction might seemingly fall outside the rules, it usually ends up being not something the FI should be concerned about. These findings are known as false positives.

What was meant to be a solution for KYC monitoring ends up being a problem for FIs. Analysts have to go through an endless list of alerts that hold no value. According to Reuters, 95% of alerts are closed as false positives in the first phase of the review. 

How does AI solve this problem? First and foremost, it moves away from the preset, rules-based system as it’s obviously too rigid and does not match the live, on-going nature of monitoring. An AI algorithm will be able to monitor transactions and determine its risk relevance before flagging it. The fun does not stop there. AI has the ability to learn from patterns, trends and previous matches and build a database where it can draw examples and references more. The more the algorithm is used, the smarter it will get, building its own set of parameters on the go.