PBSA

Except where otherwise indicated, articles are copyright © by PBSA 2024. All rights reserved. PBSA Europe PAGE 11 July 2024 AI Isn't Magic, but It Can Transform Background Checks By Alex Aoun T oday's background check industry is dichotomous at best. On the one hand, the industry has been web-enabled with new entrants offering faster online verification, and on the other, there are large swathes of the industry that still relies heavily on traditional and conventional forms of screening. The traditional or conventional approach still involves a great many manual processes, a heavy reliance on databases and a general acceptance of a "snapshot-in-time" approach to information gathering. Traditional approaches can be inflexible and not easily adaptable to detecting emerging threats. Like most other industries, background check providers cannot escape the inexorable pull of AI to modernize and automate time-consuming activities. Many reports still take days, sometimes weeks to produce, while clients are demanding faster, cheaper yet still accurate and up-to-date reports. The industry is poised for AI transformation. The twin pillars of efficiency and effectiveness AI-powered background checks can lead to greater levels of efficiency and effectiveness. AI can help maximize resources by optimizing time-consuming processes. Take, for example, the staggering amount of databases being used for virtually all kinds of checks today: criminal, credit, education, sanctions, watchlists and so on. AI algorithms can be used to search and analyze quickly, cross-referencing and verifying across disparate databases. This can save time while increasing accuracy. And, what about the information not contained in databases, such as negative news? Negative news screening or adverse media isn't new to the world of background checks. Many providers have offered these checks for more than a decade. But if ever there was a process ripe for improvement within the industry, negative news screening is it. Criminal and credit checks don't always tell the full story. Many potential business stakeholders will have impeccable credit and criminal records, but not all bad behavior is captured in databases. Negative news screening today involves the use of open- source intelligence (OSINT). OSINT is the collection of data from publicly available online sources. It's a way of classifying and analyzing the billions of historical records online and the millions more being posted every day. OSINT is referred to as unstructured data. Unlike the organized, list-based, easily searchable information found in databases; unstructured data is untamed, unmanageable, unrefined. It comes in myriad formats, such as articles, blogs, government records, online forums, social media posts, video and audio files, and so on. If efficiency is about optimization, then effectiveness is about achieving the right outcomes, which is hard to do today given the current state of negative news screening. This type of check is the bane of the screening world. It is arduous, painstaking and prone to error. Not a single analyst or researcher wants to do adverse media search, but it is becoming more important than ever, especially in regulated industries or in jurisdictions where criminal record checks are initiated by candidates. Negative news screening can serve as a first layer of checks. And, while this type of screening isn't new, many traditional data providers do a poor job of delivering relevant, meaningful, and accurate information. Today's negative news reports are rife with false positives, negatives and a lot of noise in the form of irrelevant information. Natural Language Processing (NLP), a type of AI that understands human language, is one tool used in performing negative news screening. These advanced AI models can extract, filter and analyze negative news from OSINT in real time, improving both efficiency and effectiveness, which are being supercharged by GenAI. GenAI can help to tell a fuller, more comprehensive story about a potential candidate. AI is not magic The opportunities to transform the industry are many, but AI transformation comes with its fair share of challenges. Despite the sense of wonder and mysticism that's sometimes applied to the subject these days, AI is not magic. AI is all about algorithms, datasets and data training - not magic. Continued on page 12 The industry is poised for AI transformation.

• Cover