What are the alternatives and challenges of AI within the fraud prevention and identification verification area? We caught up with Heidi Hunter, Chief Product Officer for IDology, a GBG firm, to search out out.
IDology delivers a complete suite of identification verification, AML/KYC, and fraud administration options to assist companies drive income, deter fraud, and preserve compliance. Based in 2003, IDology made its Finovate debut in 2012. GBG acquired the corporate in 2019.
Ms. Hunter joined GBG Americas in 2011 and has labored in each product innovation and buyer success roles throughout her profession with the corporate. She brings greater than 13 years’ expertise in supporting prospects and serving to them with their enterprise wants by way of product innovation, help, and implementation roles.
At the moment, Ms. Hunter is chargeable for driving the corporate’s product roadmap and bringing new improvements to the identification verification market by way of strategic product improvement.
AI has introduced on challenges and alternatives relating to fraud and monetary crime. What are the principal challenges monetary establishments are going through?
Heidi Hunter: There are 4 most important areas of concern: cybersecurity and fraud, biased fashions, human oversight, and regulatory compliance.
Deloitte has written on the rising concern of AI as a cybersecurity and fraud risk, noting that 51% of executives interviewed consider that the cybersecurity vulnerabilities of AI are a significant concern. One situation is the issue of extra and higher pretend paperwork. AI will simplify creation of passports, driver’s licenses, and ID playing cards which are nearly indistinguishable from real ones. One other situation right here is elevated artificial identification fraud. Generative AI is a productiveness device for fraudsters, creating extremely real looking artificial identities at scale.
Moreover, there may be more practical phishing and social engineering. A latest research of 1,000 choice makers discovered 37% had skilled deepfake voice fraud. And Generative AI is used to gas a surge in phishing techniques.
You additionally talked about biased fashions, human oversight, and compliance.
Hunter: The usage of AI and machine studying (ML) algorithms have come underneath scrutiny with issues over information bias, transparency, and accountability. With regard to human oversight, 88% of customers reported that they might discontinue a useful personalization service in the event that they didn’t perceive how their information could be managed.
Lack of human oversight can be a regulatory concern. AI usually lacks transparency, leaving companies uncovered after they should clarify their decisioning, which has introduced expectations of future regulation. AI-generated deepfakes are shifting quick and policymakers can’t sustain.
Can the identical expertise that’s enabling fraudsters additionally allow FIs to thwart them?
Hunter: Sure, particularly when AI is paired with human intelligence. AI advantages from consultants charged with overseeing incoming and outgoing information. A skilled fraud analyst accompanying AI-based options can catch new and established fraud tendencies. This contains novel threats that AI options on their very own could miss.
From a compliance perspective, this implies companies can supply a extra clear answer and handle potential bias. Supervised AI can get rid of the necessity to manually confirm an ID, and assist present the reason wanted for compliance and regulatory necessities.
Automation performs a significant function in AI. So does human oversight. Are you able to speak in regards to the relationship between AI and automation?
Hunter: Automation is often rule-based and follows predetermined directions, whereas AI can be taught from information and make choices based mostly on that information. In different phrases, automation software program operates on a set of predefined guidelines, whereas AI could make predictions and choices based mostly on the info it’s introduced with. The ‘predictions’ side of AI- and ML-based tech is the place human supervision performs such an necessary function.
What’s the correct stability between human oversight and AI? What function do people have in an more and more AI-powered world?
Hunter: Like with any device, human-supervised AI is nice when it’s one half of a bigger identification verification (IDV) technique.
People have a job at each ‘stage’ of AI use or implementation: in improvement, by way of what information is getting used to coach a mannequin; throughout deployment, the place an AI-based device is used and to what diploma; and relating to holding AI-based instruments accountable. This implies analyzing a given output and what choices a FI makes based mostly on that output.
For identification verification particularly, how has human-supervised AI helped clear up issues?
Hunter: Customers additionally set the bar excessive for seamless interactions. For instance, 37% of customers deserted a digital onboarding course of as a result of it was too time-consuming. Overcoming this problem requires a complete technique. Human-supervised AI can play a crucial function within the course of, as it may shortly scrutinize huge volumes of digital information to uncover patterns of suspicious exercise whereas additionally offering perception and transparency into how choices are made.
Are companies embracing human-supervised AI? What hurdles stay to broader adoption?
Hunter: Sure, as a result of whereas there may be numerous pleasure round what AI can do, a number of companies and folks within the educational neighborhood consider AI isn’t able to make unsupervised choices. As talked about earlier, companies present concern over AI working by itself. Considerations vary from moral questions, to cybersecurity and fraud dangers, to creating a nasty enterprise choice based mostly on AI. On a constructive notice, companies have gotten extra conscious of advantages of supervised studying fashions.
Photograph by cottonbro studio
