September 4, 2019
Eliminate the Noise: How AI Contributes to Fighting Fraud
Posted by Laine Sulik
The Association of Certified Fraud Examiners (ACFE) partnered with SAS to publish the 2019 Anti-Fraud Technology Benchmarking Report which predicts that “the use of artificial intelligence and machine learning as a part of organizations’ anti-fraud programs is expected to almost triple over the next two years.” As technology continues to evolve, so do the criminals in their development of fraud schemes. As the ACFE reports, “technological advancements present opportunities for both fraud perpetrators and those trying to stop them.” Organizations that incorporate artificial intelligence (AI) and machine learning into anti-fraud programs will be better equipped to detect and deter fraud.
Where Can AI Help?
There is a vast array of AI tools on the market today. In this article, we outline the most relevant to enhance your fraud prevention programs.
Artificial Intelligence (AI) and machine learning can scour through colossal data sets, or “Big Data,” to identify irregular or suspicious transactions in a fraction of the time it would take to apply traditional data analysis techniques. Big Data could mean petabytes or exabytes of data, billions to trillions of records, volumes of information that is unfathomable to some. According to the ACFE, 64% of companies stated that data analytics were very beneficial in analyzing large volumes of data and 59% of companies stated analytics were very beneficial in allowing them to detect anomalies more quickly. With AI, companies can train their data analytics programs to recognize red flags in real time as transactions occur.
Data collection and conversion
Companies collect data from countless sources, both internally and externally. In many instances, internal data is not sufficient and must be aggregated with external data to provide valuable information for decision makers (public records, government watch lists, social media, third-party data, emails, images, etc.). Data engineers who are well-versed in computer science and have a strong understanding of information systems typically complete this process. AI can help experts navigate the process of linking information from various departments or systems.
Once compiled, Data Scientists clean and analyze the data to understand any underlying relationships between the variables. Specialized algorithms are then applied to extract business insights from the compiled data. A significant portion of data analysis work involves collection and conversion. Data Scientists are instrumental in determining what to do with missing data points and how to best leverage the available algorithms to produce accurate and reliable information.
Balancing Human Error
We must not discount the fact that there will always be a human element to fraud detection. However, AI is changing the way we approach fraud prevention and detection in a significant way. Despite what the media portrays, AI is still a tool. And tools are designed for a specific purpose. Those that are used incorrectly can cause more harm than good. AI is only as good as the team behind its design. Understanding the data feeding the algorithm is the most important aspect of developing a robust and useful model. The best AI tools will still require human interaction; remember, these tools are ARTIFICIALLY intelligent. If humans are fallible, then the “intelligence” designed by humans is also fallible. Monitoring AI tools for efficacy over time is not only smart—it is a protection of your investment.
Because we have access to advanced technology, companies can now focus on the prevention and deterrence of fraud rather than having to jump into crisis mode after a fraud occurs. The ACFE’s Report to the Nations: 2018 Global Study on Occupational Fraud and Abuse emphasizes the value of preemptive fraud detection methods as “fraudsters tend to start small and increase their frauds rapidly over the first three years.” AI can point to suspicious data where companies can apply resources for further investigation, flagging items that show certain keywords used in emails, transaction activity outside of regular business hours or on weekends, or whether the address of a new vendor is zoned as residential. Early identification of these and other threats mitigates the potential fraud risk and can significantly reduce the window available for fraudulent activity to occur. AI can help organizations tailor their analytics to produce results that are truly meaningful rather than spitting out reports that look nice but don’t show the story behind the picture.
The ACFE’s Report to the Nations also reports that “frauds that last over 60 months are more than 20 times as costly as those that are caught in the first six months.”
AI recognizes areas of risk by training itself to identify circumstances that are susceptible to fraud, ultimately leading to early detection. Machine learning algorithms look for periodic trends or similarities in historic data that can then be leveraged when other data fits similar patterns. Business insights are improved when additional information is added to the data set. Data Scientists rely on feature engineering to construct new fields that improve machine learning efficacy.
Layering Data Analytics
With the use of AI, organizations can generate heat maps showing the geographical locations of vendors, and at the same time, we can see which of those locations are zoned as residential. We can see the number of transactions per vendor each month and simultaneously see what percentage of those transactions fall within the last week of the month. AI can save us hours and hours of filtering and re-running reports to then spend more time cross-referencing data to drill down to the results we need for informed decision making.
Benefits of Utilizing AI
Fraudulent activity is considered anomalous because it falls outside the “norms” of typical transactions. As such, machine learning algorithms are perfectly suited for recognizing these financial outliers. When used properly, Artificial Intelligence can reduce human errors and save countless hours of analysis by automating time consuming, manual processes. Ultimately, this allows employees to focus their efforts on value-adding, strategic activities rather than mundane, manual tasks.
The Armanino AI Lab
The ACFE reported 80% of companies had concerns around the financial implications of implementing AI procedures into their anti-fraud programs. The Armanino AI Lab helps our clients harness the opportunities that AI offers without being left behind by their competitors. It is the perfect opportunity for companies to participate in peer-to-peer meetings; have exclusive access to product and vendor overviews; connect with Armanino’s data scientists, AI developers and consultants; learn about AI best practices; and execute on AI proof of concepts. To learn more about the Armanino AI Lab, visit: learn.armaninollp.com/artificial-intelligence-ai-lab-signup/.
Laine is a Certified Fraud Examiner and analyst who has worked on forensics and due diligence projects for clients across many industries. She has extensive experience analyzing financial data, reconstructing missing records, locating hidden assets, calculating economic damages, and preparing expert witnesses for trial. She works closely with legal teams during discovery and throughout litigation, serving as liaison between clients and agencies such as the FBI, IRS and district attorney’s office. She also has experience assisting clients with financial due diligence analysis.
Laine earned both a Master of Science in accounting and a Master of Business Administration from the University of Dallas; she also holds a bachelor’s degree in psychology from Texas A&M University.
Co Authors :
|After graduating with BS degrees in Chemistry and Biochemistry, Derrick worked as a researcher where he focused on improving a number of advanced scientific instruments. After spending a few years in a lab coat, Derrick went on to attain an MBA and a Master’s in Electrical Engineering where he specialized in Control System Theory. Over the last two years, Derrick has leveraged this diverse skillset and unique perspective to teach organizations how to capitalize on their data to develop sustainable, competitive advantage.|