Accounting

The Future of Accounting and Auditing is AI, Bots, and Machine Learning

Jul 6, 2021 Brandon Malekie

No, the robots are not going to take your job in accounting, but your new co-worker just might be a bot. By 2025, economists at the World Economic Forum project that the time spent on tasks by humans and machines will be equal. 

Bringing artificial intelligence, or AI, into accounting and audit will make that work more fun and provide the deeper insights that businesses crave. For the first time in the history of accounting, accountants will get a break from the boring tasks that bots and AI can do faster and more accurately, and they’ll get to do more of the work that actually requires a CPA. That’s happening today in accounting firms and in accounting teams in industry that are implementing AI technology. Let’s take a look at what this explosion of artificial intelligence can do for accountants and auditors. 

Artificial Intelligence In Accounting And Auditing

First of all, what is artificial intelligence? A broad and general definition comes from CPA Canada, which defines AI as “the science of teaching programs and machines to complete tasks that normally require human intelligence.” Artificial intelligence uses computer algorithms to perform tasks, offer suggestions, and even make decisions in ways that imitate and complement human intellect. 

At its most basic level, artificial intelligence is task-based with limited capabilities: a certain input results in a specific automated response. For example, an accounting system can be taught to automatically code transactions from a bank feed based on rules provided by a user. Robotic process automation, or RPA, uses AI to automate certain tasks, such as pulling an order from a customer portal, extracting the data, creating an entry to accounts receivable, and routing the information to order fulfillment. At this level, what the AI bot does is limited to the user-defined rules.

But today’s AI has advanced beyond simple rote memorization of tasks to learn on its own. Machine learning, or ML, uses AI algorithms applied to large datasets to identify patterns and make predictions or decisions with limited human guidance. This type of AI gets better as it processes more data, and may apply natural language processing to interpret spoken or written language. Recommendations from Amazon or Netflix are examples of this kind of AI. The datasets here may be structured, such as a set of accounting transactions, or they may consist of unstructured data, which can include satellite images, email messages, and audio or video files — basically anything under the sun that can be digitized. 

Deep learning is a subset of machine learning that models the human brain. It uses artificial neural networks where each layer in a hierarchical structure passes on what it has learned to the next layer. Deep learning applies nonlinear machine learning algorithms to identify and learn from very large datasets. Self-driving cars use deep learning to interpret the many sensory signals that we humans can process and respond to instantly. 

We all use AI in our everyday lives. Our social media newsfeeds are curated by algorithms that give us more of what we’ve already given our attention to. Support on many tech websites relies on programmable chatbots using natural language processing to provide solutions to the most common problems. Our banks use AI to alert us to potentially fraudulent activities. In the healthcare field, AI is being used to help radiologists analyze mammograms to detect cancer.  

While ideas for implementing AI technology in accounting and auditing have been around for decades, it was not until recently when the combination of faster computers, cheap data storage, and the increasing digitization of business transactions made the development of AI tools commercially viable. Today, we see an explosion of emerging technologies in all sectors. 

The Three Types of AI

AI comes in three types, depending on the level of intelligence embedded in the system. The most primitive — which at present, is all we’ve developed so far — is Narrow or Weak AI. This type of AI is task- or goal- oriented. This is the AI in IBM’s Watson, which is good at specific tasks, like winning Jeopardy.  

Next up is General or Strong AI, which seeks to be as intelligent as a human. This is the AI of HAL in 2001: A Space Odyssey. The highest level of AI is Artificial Superintelligence, which surpasses humans in all aspects of intelligence or creativity. This is the AI in the Terminator series, where machines ultimately decided they no longer needed humans.  

Can Artificial Intelligence Be Used in Auditing and Accounting?

Because bots never get bored and perform their tasks flawlessly and quickly, AI is especially useful for the repetitive, error-prone manual processes that have defined the work of accountants and auditors for centuries. Now, thanks to new technologies being implemented by forward-thinking accountancy firms and accounting professionals in industry, accountants will have the time and energy to provide the higher-value consultative work that actually helps organizations achieve their goals. 

Beyond simply following user-provided rules to code transactions or to automatically perform specific tasks, today’s machine learning powered AI can provide real-time insights that provide a competitive advantage above organizations still relying on manual processes.

Here at FloQast, we use AI in AutoRec to easily match banking transactions to GL transactions. Faster reconciliation helps close the books faster, which means that the financial statements and reports used in decision making get to organization leaders sooner. 

Over in the accounts payable arena, besides using optical character recognition to interpret and correctly code and record a vendor invoice, apps using AI may be able to perform these tasks:

  • Match the invoice to goods or services delivered and to purchase orders 
  • Extract information from accompanying contracts or other documents
  • Send alerts about discounts for early payment
  • Route the invoice to the correct person for approval
  • Forecast cash balances to determine the best time for payment 
  • Identify human-caused errors
  • Eliminate duplicate payments
  • Detect fraudulent AP schemes
  • Reconcile payments to invoices 

By automating these time-consuming and labor-intensive business processes, accounting teams can do more with a smaller headcount. In today’s complex regulatory environment, where staffing has been a challenge for years, hiring an AI bot to do the time-consuming and repetitive tasks can be a true game-changer. 

In the wake of the pandemic, many organizations are developing multiple-scenario financial forecasting models to help them respond to sudden changes in their operating environment. Adding machine learning to forecasting allows FP&A teams to include vastly more data from many more sources to increase the accuracy of their projections. These kinds of projections are far beyond the power of Excel. 

Over on the audit side, AI tools assist with risk assessment by combing through the entire GL to identify questionable transactions. Instead of sampling just a few transactions, and hoping to find the needles in the haystack, audit teams can test 100% of transactions and find all the needles, thus considerably reducing audit risk. 

How Can AI Improve Audit?

Machine learning has the potential to increase the speed and quality of audits, as the  CPA Journal explains here. By using AI for the tedious ticking and tying tasks that are an inevitable part of the auditing process, auditors have more time for review and analysis, and are more able to focus on the more difficult, higher risk areas. The extra time and mental energy gained provides auditors the ability to step back and see the big picture. Audit firms that leverage AI along with data analytics to take a data-driven approach to audit will have a competitive advantage because they can provide valuable insights to their clients beyond the audit report. 

Some tools use AI to test journal entries to find questionable entries. Others correlate financial data with unstructured nonfinancial data to confirm that business performance is consistent with its operating environment and business model. These tools aren’t a substitute for an auditor’s understanding of businesses and industries, but can augment that auditor’s human intelligence to provide valuable insights to clients that will help them achieve their goals. 

In the not-too-distant future, audit firms may be using bots for routine tasks such as confirmations and to assist in straightforward audits such as employee benefit plan audits. 

Be the Disruptors, Not the Disrupted

While it is true that AI, machine learning, and robotic process automation have the potential to  take over many of the tasks that accountants and auditors perform, they won’t replace the professional judgment, insight, and guidance that only a human can provide, at least for the foreseeable future. 

Technological disruption is on its way, and it will transform every part of the accounting profession, and has the potential to disrupt accountants out of their jobs unless they become part of that transformation. Audit firms and accounting teams in industry that implement these new technologies will help ensure that the accounting profession remains relevant and won’t be replaced by evil computers like HAL.