Tom Hood has seen the audit industry go through plenty of changes over the years, but the president and CEO of the Maryland Association of CPAs says that nothing may alter the sector more than artificial intelligence (AI). Over the past year or so, an increasing number of firms have begun to adopt AI-enabled tools that allow their CPAs to do their work in far more detail and with greater speed than ever before. “We’re in the fourth industrial revolution,” says Hood about AI. “The scale and the speed of change is really starting to challenge our profession.”
While some people may see AI as a threat, Hood sees it as an opportunity. Technology will allow CPAs to improve their accuracy, identify issues they may not have been able to see before and free up their time from mundane tasks to provide companies with more business-building advice. AI can be especially useful when identifying anomalies or errors, such as a credit balance in accounts payable, or a flurry of activity in a given time period. With the tremendous growth of data in the past several years, it can be challenging for humans to identify and catch anomalies. For instance, one sign of fraud is when someone inputs an uncommon entry of an unusual amount in the middle of the night ie an outlier transaction. That can be hard for humans to catch with the multitude of entries over a given time period, but computers can look at every single entry and flag the ones that are suspicious.
Finding unusual transactions with AI
One program in particular has changed the way some of Hood’s member firms work, especially when it comes to finding unusual journal entries. As these entries could either be an indicator of fraud or unintentional errors, they may require further investigation or analysis. Ottawa’s MindBridge Analytics looks at all transactions to determine whether something nefarious has taken place or if something has been inputted in error. It might notice that several payments have been made to one address over a short period of time and flag that for further investigation, or it can tell an auditor if numbers don’t add up correctly.
To find unusual transactions, and potential cases of fraud unintentional errors and anomalies auditors apply traditional methods, which usually involves taking a random sample of a client’s financial records—for example a general ledger—to determine if there are any anomalies. With MindBridge, audit firms can look at all transaction/entries and identify any high-risk items.
“Imagine doing 100 per cent almost instantly,” says Hood. “That’s significant because it is providing a completeness check that has not been available previously, and it’s changing the role of the auditor and automating a big chunk of what is done by humans.”
A more powerful auditor
The impetus for starting MindBridge was that the current analytic tools were only detecting three per cent of fraud cases. Most fraud is uncovered by whistleblowers or by chance; using AI to look at all of a company’s financial data is a huge advantage in the fight against fraud, as it provides a higher level of assurance and reduces the risk of fraud.
“There’s no company in the world that would employ the amount of staff it would take to do that kind of check of the books,” says John Colthart, MindBridge’s vice president of growth. “Now you can present all potential anomalies in a matter of minutes.”
The company’s software “gives the power back to auditors, accountants and CPAs,” says Colthart, in part because it’s so easy to use. The MindBridge’s AI platform assists in automating the ingestion of data, and it can start reading files right away. Auditors can also set a time frame to explore—even longer than a year—to see if any discrepancies have occurred in the past. The results come back quickly, at which point the auditors can start going through all the issues the AI has identified.
“Investigations that have historically involved months of combing through millions of transaction samples are now much more comprehensive—yet more narrowly focused—and can be concentrated in just hours,” said Kirstie Tiernan of BDO USA, which joined in a strategic partnership with MindBridge in July 2017.
A learning machine
The program works, because MindBridge applies a hybrid of algorithms against every transaction; it doesn’t rely on a single technique. These algorithms, which exceed Generally Accepted Audit Standards, include models based upon business rules, statistics, machine learning and artificial intelligence. The AI also learns on its own—if something gets flagged that’s not an issue for a firm, then it’ll know not to highlight it next time. “As people interact with the Ai Auditor, it continually learns,” says Colthart.
Glass Box approach to AI
MindBridge has spent considerable time and effort to explain why certain transactions were flagged to avoid potential “black-box” concerns.
“If someone asks me why we have audited a particular sample,” says Becky Shields, partner at U.K. accountancy firm Kingston Smith, which has partnered with MindBridge, “I can explain the computer-based technique, which is a lot more robust than saying, ‘One of my trainees picked 10 transactions’.”
It will still be a while before AI is used by every audit firm—Colthart says that while MindBridge are already making a difference today, it is continually improving its AI every day by feeding it new data points. One of Hood’s member firms, which uses MindBridge, says it has given them new insights into clients they could never get before, while Colthart says the software has helped identify several frauds.
At some point, though, AI will be everywhere and firms that adopt it early will be the ones that get ahead. “We’re going to see AI use in auditing double at least every year for the foreseeable future,” says Hood. “Everyone needs to know more about it.”
Want to learn more?
John Colthart and Tom Hood will be speaking at The ONE Conference on Oct. 1-2. Hood will also be discussing “Skills for the Future” at the 2018 AI In Accounting and Auditing Conference in Maryland in December.