Audit professionals are using artificial intelligence in ways that are both expected and surprising, categorizing risk across entire ledgers and feeder systems without requiring a person to pore through vast amounts of data.
Beyond the sheer processing power of AI, it enables a new evolution of reasonable assurance. AI learns about the data it’s analyzing, determining patterns, and identifying outliers over time. As it learns more about the deep financial details and history of clients, it can quantify and flag behavior or outliers to auditors that it considers risky.
While academics and some industry professionals might say these benefits are theoretical, the reality is that they are available to auditors right now.
What AI Does for Auditors
A common example of how artificial intelligence algorithms are applied to audit is the detection of material misstatements using “unsupervised learning.” These techniques leverage the science of determining what is usual versus unusual to report on outliers in ledger data without bias or history, letting the data speak for itself.
K·Coe Isom, a leading consulting and accounting firm for the food and agriculture business, uses AI to provide unique insights and a complete view on financial health for clients. Brittany Ferguson, Senior Associate at K·Coe, explains, “We used AI-based analysis for materiality limits and extracted medium and high-risk items to run samples on during our planning stage. This risk assessment identified two transactions that would not have been found under traditional testing conditions. The finding, although immaterial, was a value-added training opportunity that we were able to offer to the client.”
AI warrants a re-evaluation of how audit planning and testing are done. Historically, the only feasible method for substantively testing vast quantities of data was to sample transactions statistically or non-statistically, foregoing the effort necessary to examine the entire dataset. This often required substantial back and forth time with the client to obtain the requisite information not received during fieldwork.
“A slowdown in any audit is having to wait on client responses to questions," says John Downey, Senior Audit Associate with Doeren Mayhew, a Top 100 firm with offices in the U.S. and Europe. "With artificial intelligence, both us and the client are more efficient during the process because we have all the data already. With one client, there were year-end journal entries in inventory and accounts payable posted by the controller that we wanted to investigate. In the past, we would have to go back and ask but this time, we just pulled up the details right away.”
How AI Redefines Fraud Investigation
According to the 2018 ACFE Report to Nations, two out of the top three categories of occupational fraud are asset misappropriation and financial statement fraud. As both methods can involve different forms of records manipulation or report tampering, AI is the ideal investigative tool as it can uncover anomalous activities across 100% of the data.
A recent fraud case with a consumer goods manufacturer provides an example. As explained at ACFE Insights, the California audit services firm of Gursey Schneider LLP used AI to analyze more than $2.8 million in fraudulent transactions. Gary Krausz, CPA, CPF, explains that “Once you’ve exceeded the capabilities of Microsoft Excel or traditional CAAT tools, your auditor quickly runs into a wall.” AI was able to ingest and understand all of the data, reporting on transaction items of unusual amounts and accounts that warranted further investigation by an auditor.
Increasing Public Trust with AI
No matter the country, the public’s trust of governmental budgeting, fiscal management, and reporting is at an all-time low. According to the OECD Trust and Public Policy report, “The global financial crisis blew public finances off course...This has severely damaged fiscal contracts and resulted in citizens having reduced trust in the ability of governments to manage public finances in a sensible and sustainable way.”
Instances of fraud, collusion, and ineffective audit and oversight practices are never far from the headlines and public servants are under increasing scrutiny to do better, often with less resources. With AI, auditors, finance and oversight officials at all government levels can understand, test and report on massive amounts of data to meet their fiduciary responsibilities to taxpayers more than ever before.
The firm CohnReznick is [in our view] at the forefront of deploying AI and machine learning technologies as a competitive advantage in the safeguarding of public funds for federal, state, and local agencies. "We’re at a unique crossroads where emerging technologies such as AI are helping government agencies increase efficiency, improve the job satisfaction of public servants, increase the overall quality of services offered, and help them advance their mission more effectively,” says Bill Hughes, Partner, Federal Market Leader, Government and Public Sector at CohnReznick. “We’re seeing improvements in public trust and fiscal responsibility at all levels of government as we move from looking at small samples to analyzing data sets so large, they could not be tested without AI."
AI for audit is real and in use, helping firms utilize all the available data with improved speed, risk assurance, and meaningful insights for clients. It’s crucial for business leaders to select the solution that allows their firm to shift from standard audit reporting and compliance to value-added advisory services that offer competitive differentiation.
The question now is how to sift through the noise to best position their firm for success.
MindBridge Ai is restoring confidence in financial data with Ai Auditor, the world’s only AI-powered auditing solution that leverages machine learning and AI techniques to augment human capacity and redefine reasonable risk assurance.
John Colthart is General Manager, Audit and Assurance & VP Product Management at MindBridge Ai. After his departure as a corporate finance and accounting practitioner in 2000, he grew a startup to over 425 employees and exited to IBM with a role of VP Sales Operations in 2010.