How Machine Learning is Disrupting Accounting
In the accounting, audit, and compliance professions, Artificial Intelligence(AI) is already beginning to automate labor-intensive tasks, such as data entry or combing through manual documents.
For example, accounting and audit professionals may deploy AI to extract information from invoices or purchase orders to enter into accounting and auditing systems, freeing up hours in the day to perform more robust analytics on the outputs of the system. Taking advantage of AI could enable accounting, audit, and compliance departments to analyze significant amounts of data, and deliver more analysis and insight as a part of their roles.
In fact, AI has been making our lives easier for longer than we may realize. Your smartphone, your car, your bank, and the devices in your house may use AI on a daily basis to predict your preferences or anticipate your actions.
AI is the discipline of training machines to make human-like decisions and perform “smart” tasks that normally require human intelligence. The recent interest in AI has been fueled by a trifecta of advances in machine learning techniques, ever-increasing data availability and continued acceleration in computer processing. AI developers have created applications that can augment human performance, automate complex processes, and simulate human engagement to increasingly sophisticated levels.
Data is a big opportunity…and challenge for AI
AI is only as intelligent as the data that feeds it. To deliver on the potential of AI, machines “learn” from large repositories of information. The process of training a machine requires brute force computation – the equivalent of a teacher correcting students “yes” or “no” as they try to answer questions in rapid fire, until they have (almost) perfectly memorized all the answers to the questions.
While data is empowering AI and machine learning at scale, getting access to quality data sets to solve specific business problems remains a huge challenge. Unstructured data – whether it’s text, images, or audio – must be digitized and transformed into a source of “ground truth” before AI-powered solutions can be created. This data wrangling exercise can be daunting for many organizations.
Fortunately, if organizations lack the information or skillset needed to train their own AI systems, increasingly they may consider pre-trained systems that accomplish specific cognitive tasks. Many businesses have use cases for AI that are already trained for speech recognition, natural language processing, or computer vision. These systems may even be used to digitize and automate existing processes, opening those processes up to further AI development.
Machine-human collaboration is crucial
By no means does the introduction of AI into accounting, audit, or compliance signal the end of the human professional. There will always be a place for humans to exercise professional judgment in their work – in fact, AI is an enabler.
While machine learning algorithms have steadily advanced toward simulating business processes, they primarily classify information. They lack the context to reach novel conclusions.
Most AI solutions are inherently single-purpose and can be brittle. Back to our student analogy, you can imagine that a student trained on a single subject matter may have impressive knowledge on that topic.
However, the same student may be unable to connect the dots beyond that area of experience. For example, an AI system that is built to extract data from purchase orders will not spontaneously begin recommending procurement improvements.
It may reach levels of accuracy that exceed a human performing the same routine tasks, but would be unable to advance into other areas, much less higher level thinking. This is where humans need to embrace machine-enabled decision-making.
Simply put, humans will not win a data processing competition with machines. At the same time, AI systems are not adept at self-correction in new circumstances. When AI misses, it often misses badly.
In contrast, humans bring context to their decisions, quickly integrating new information into a complex knowledge framework. Accordingly, AI systems in their current state should be considered extensions of their users, and more akin to a tool than a competitor. After all, many AI systems are relatively “young”, having received intensive training for only a few weeks or months before being placed in service.
AI is an opportunity for audit and compliance professionals to “lean-in”
As AI is being used to find the proverbial needle in haystacks of data, auditors, compliance professionals, and accountants should challenge themselves to identify more rewarding problem-solving activities. For example, in the future, auditors may find that using AI to automate routine procedures allows them to focus on extending assurance to other, non-financial statement information in the same time frame.
Likewise, accountants and compliance professionals, relieved from potential low-value data extraction and preparation activities, may be able to take advantage of their institutional knowledge and focus on business process improvement. Compliance professionals and auditors should also consider their role in AI governance.
Adopting a new technology like AI introduces new risks into an organization. Because AI systems may have difficulty in the face of change or with deliberate attempts to mislead it, businesses will need to monitor their AI systems’ performance.
Similarly, if accountants begin to use AI to process information for the preparation of financial statements, the auditing profession will likely need to develop standards to evaluate the reliability of these black-box-like systems.
There is also vibrant debate within academic circles related to the ethics of AI. Since AI is trained on historical data, it can inherit the bias embedded in the “ground truth” data.
Given the maturity of AI, audit and compliance professionals should be aware of what is being “delegated” to AI within specific business processes, along with the potential risks. Auditors and compliance professionals may find themselves with the opportunity to help oversee, or “audit” the complex algorithms that guide AI systems.
What’s next for AI in accounting, compliance, and audit?
The use of AI across business functions like accounting, compliance, and audit can be expected to enable professionals’ ability to analyze and act on relevant information, with less data processing distraction. Furthermore, auditors and compliance professionals have an important role in identifying and monitoring risks introduced by the proliferation of AI in business processes.
We should begin thinking about AI as an enabling tool – rather than a direct competitor. We should use these tools to leverage our precious time, improving our job satisfaction. At the same time, we should be mindful that some of this new time needs to be dedicated to also watching the machine.