TECHNOLOGY INSIGHT
AI and expenses fraud
Adam Reynolds, chief executive officer of webexpenses, explains how AI can help to fight such fraud
I t’s nearly twenty years since world chess champion Garry Kasparov was first beaten by a computer opponent – the Deep Blue system developed by IBM. Since that historic event, artificial intelligence (AI) has started to play an increasing role in our world. One area, in particular, where a computer’s ability to learn the best way to complete a task has found a natural home, is in the financial services sector. A recent study by Thomson Reuters found that 75 per cent of all global trades are now AI based transactions. The AI systems allow finance managers to make sense of the chaotic fluctuations of a financial market; analysing vast quantities of data to identify patterns and trends that inform the decision-making process. Until recently, the technology required to power these kind of AI systems was prohibitively expensive with high-end computers required to handle the amount of number crunching required by complex algorithms. But as computing power increases, this technology is starting to filter its way down into our everyday lives. For payroll professionals, it has considerable potential to become a powerful tool, helping to unlock a smarter way of managing company costs. It would allow the data generated by a digital expenses management system to be automatically analysed, searching for any suspicious spending patterns. Through the process of ‘machine learning’, the more company data it received, the more accurate it could become at spotting risks.
The power of AI to identify fraudulent behaviour is something that’s already successfully used by credit card companies such as MasterCard and RBS WorldPay. They have developed systems which track card usage and flag any suspicious transactions. ...AI would open up the strange possibility of predicting problems before they have even happened They determine the risk of each transaction by looking at everything from its geographic location to the credit card owner’s typical spending patterns. The real-time nature of the systems mean that suspicious transactions can be instantly blocked, until validated by the card owner. Elements of this have already found their way into digital expenses management. A system such as webexpenses can be set-up to automatically check each claim made against an organisation’s expenses policy. If a claim falls outside of policy limits, it triggers a real-time alert. This can send a notification to the claimant and allow a finance team to make checks that it’s a valid business cost. With AI, there is the potential to extend this kind of feature to provide a more nuanced approach to sniffing out fraud. By
analysing expenses data it could start to learn the patterns and trends for a specific business. This information could be used to build up profiles for particular job roles and search for areas where claims fall outside the norm. It could identify those traits which are most commonly displayed by people who falsify or exaggerate claims. It would be able to highlight risks even when a fraudster was trying to cover their tracks by staying within the acceptable policy limits. And the more data it crunched, the smarter it could become at identifying these areas of concern. While current systems are able to block a suspicious claim as soon as it’s received, the use of AI would open up the strange possibility of predicting problems before they have even happened. It could identify those areas of an organisation where the risks of fraud taking place are ‘hot’ or ‘cold’, allowing preventative measures to be carried out. This may be addressing underlying problems or boosting compliance procedures. So, for any finance professional tasked with managing company costs, AI presents some intriguing future possibilities. It would move expenses management further along its current trajectory, moving from a reactive to a preventative approach. But no matter how complex and advanced the algorithms become, an AI system will be of little use without an effective finance team behind it. It’s a tool with the potential to change the way finance professionals work but not to replace them. n
| Professional in Payroll, Pensions and Reward | December 2016/January 2017 | Issue 26 46
Made with FlippingBook Online document