Beyond robotics: How AI can help improve the audit process

The CPA profession has been hearing a lot about Robotic Process Automation (RPA), a software technology that can help auditors sift through structured data. But RPA can only do so much. Abigail Zhang, a PhD student in the Department of Accounting and Information Systems at Rutgers Business School, explains why auditors should also consider Intelligent Process Automation (IPA).

What is IPA?

Intelligent Process Automation (IPA) includes:

Robotic Process Automation (RPA)

Artificial Intelligence (AI)

Cognitive Computing (CC)

Other technologies

While RPA is good for some tasks, IPA expands its functions to make the automation scalable, flexible and intelligent. IPA is the trend of the future in automation, and the major RPA vendors are competing to make their products AI-powered.

What are the limits of Robotic Process Automation (RPA)?

RPA is “robotic” in the sense that it can only automate tasks that are repetitive, standardized, structured and rule-based. It cannot adapt to changes. RPA cannot:

Deal with unstructured data (e.g., emails, social media postings, news and audio) or semi-structured data (e.g., invoices and contracts).

Make complex decisions that require a higher level of cognitive judgements.

Analyze and further process exceptions.

What makes Intelligent Automation Process (IPA) preferable for audits?

In general, IPA can do whatever RPA can do and then some.

IPA integrates artificial intelligence and other technologies with RPA, unlocking the potential in each technology. IPA forms an intelligent digital labor force that can help humans with tasks that RPA alone cannot handle. Besides the RPA-type structured tasks, IPA can also process unstructured data like emails, perform complex data analysis, process exceptions, conduct predictive analysis, adapt to changes and learn through time.

Unlike RPA, which can only execute pre-programed procedures, IPA can “sense,” “think” and “act.” When the IPA is not able to deal with certain tasks, it will forward them to a human and learn from what the human does.

IPA can potentially be implemented with many repetitive knowledge workers’ tasks. For example, in the banking industry, IPA could be used to understand and act upon digitized documents, optimize solutions to unexpected events, monitor human transactions and prompt alerts for any abnormal activities. In the finance industry, IPA could be used in account reconciliations, financial closures, and financial planning and analysis. In the insurance industry, IPA could be used in underwriting and claims/billing processes.

What can’t IPA do?

IPA can’t totally replace human workers. Although IPA contains artificial intelligence and cognitive technologies, it will not – and is not supposed to – expel humans from the workflow. Rather, IPA is aimed to diminish employee time spent on manual and repetitive tasks, to assist them in creative and high value-added tasks and to help make better decisions.

How can auditors implement IPA?

The general rule for implementing IPA is to “start small and then grow.” Firms, regardless of size and industry, should start with RPA to automate the tasks that are most time-consuming, repetitive, standardized and rule-based. These are the “low-hanging fruits” that can immediately benefit firms. Firms can then gradually add artificial intelligence and cognitive components.

When adopting IPA, firms should consider the following questions:

Is the task structured, semi-structured or unstructured?

Is the task automatable?

If the task will be automated, what potential automation tools can be used for this task?

Is there any need to redesign the process for more efficient automation?

How can small CPA firms benefit from IPA?

IPA might help small firms remain competitive in the market. First, because it operates on top of existing firm systems, the IPA solution is generally more affordable and agile than creating new IT systems. Second, if properly implemented, the time and cost savings from the tasks that are most repetitive, standardized and rule-based will be immediate and very large. Third, if the firm uses a vendor-supplied RPA/IPA platform, they can avoid having to hire a “hard coding” expert. Firms would be able to directly use the artificial intelligence and cognitive functions in those platforms right off the shelf.

How might clients and the public benefit from IPA used by auditors?

The assurance profession can also benefit from IPA. Currently, auditors are still burdened with high-volume and repetitive tasks. In my research exploring the potential applications of IPA in audits, an “auditor-in-the-loop” IPA ecosystem was constructed. In this ecosystem, the structured and semi-structured tasks (e.g., reconciliations, recalculations and confirmations) can be automated in an unattended way, almost without auditors’ intervention. The unstructured tasks (e.g., internal control risk assessment) can be partially automated (i.e., attended automation) in the sense that AI/cognitive computing will assist auditors. In the IPA ecosystem, auditors and bots will augment each other’s abilities and cooperate to complete audit engagements.

Also, there could be a major benefit from IPA in the audit: increased audit quality. First, the auditors can focus more on the tasks that need professional judgment, and they can make better decisions with the assistance of AI and cognitive computing. Second, automation can almost eliminate human errors in repetitive tasks. Third, IPA allows auditors to perform full population testing, which can provide more convictive support for the audit opinion than traditional sampling methods.

Any final thoughts?

“Automation is constantly changing and does not stop at RPA.” Accounting professionals should be prepared for and have an open mindset about the future.

The information in this blog is based on the working paper, “Intelligent Process Automation in Audit” by C. Zhang, 2018.

Chanyuan (Abigail) Zhang, PhD Candidate – Rutgers Business School. Chanyuan is a PhD candidate in the Department of Accounting and Information Systems at Rutgers Business School.

Beyond robotics: How AI can help improve the audit process

Related Stories