With the growth of computing power and big data over the last two decades, the direction of artificial intelligence (AI) adoption has been forward, and it is moving fast. AI is technology that aims to boost the capability of machines to mimic human cognitive skills, such as thinking, talking, seeing, hearing or reading. Self-driving cars are a real-life application of AI. In recent years, enterprises have been investing heavily in AI. In the United States alone, investment in AI has grown exponentially, from US$300 million in 2011 to approximately US$16.5 billion in 2019.1 However, the question is how these times of economic and social change will impact AI adoption.
Within the profession of corporate compliance, auditing firms have been reinventing themselves as technology hubs, and the race to develop solutions has accelerated over the last five years. For example, during this period, the Big Four accounting firms (i.e., Deloitte, EY, KPMG, PricewaterhouseCoopers [PwC]) have been developing AI systems for their auditing, advisory and assurance functions. For instance, within the assurance practice, the Big Four firms are piloting the use of AI to “perform auditing and accounting procedures such as review of general ledgers, tax compliance, preparing work-papers, data analytics, expense compliance, fraud detection, and decision-making.”2 With fewer capital resources available for AI, small-to-medium-sized audit firms are adopting AI Software as a Service (SaaS) approaches.3
The pressure to innovate has not only come from peers but clients as well. Audit clients are implementing technologies such as the Internet of Things (IoT), which can result in fully automated, backend operations. Auditors can no longer use traditional, manual methods for auditing operations in an automated environment. For example, when a process is driven by AI decisions, who does the auditor query? Will auditors be able to safely carry out manual inventory counts in warehouses that are fully automated with robots as anticipated in a new, socially distant age?
In this new decade, a new spin has been thrown into this technological evolution. Three questions quickly come to mind when considering what uncertainty means for AI technology adoption:
- Will there be new situations that arise from the current economic and social changes that will impact the growth of AI within the profession?
- Will economic changes increase the wedge between small and big audit firms in terms of AI adoption?
- Will social changes impact the nature and future of work, giving rise to a human workforce that is digitally connected and subsequently augmented by AI?
The Impact of COVID-19 on the Audit Profession
Regulators and professional bodies have observed the growing strain on the auditing profession as a result of COVID-19. For example, the American Institute of Certified Public Accountants (AICPA) has released guidance on audit matters and reporting issues as a result of COVID-19.4 Despite government-recommended lockdowns and social-distancing guidelines, auditors still need to comply with audit regulations. One of the significant challenges posed to auditors is the ability to comply with standards, such as FASB ASC 275, that require auditors to consider the effect that changes, risk and uncertainties pose to financial statements. Beyond reporting, the ongoing pandemic pose a risk to organizations impacted by lengthy lockdowns and a shift in social behaviors. The end of the pandemic remains unpredictable and, therefore, expressing audit opinions becomes more trying. In addition, the auditor’s role extends beyond the reporting date since auditors need to report on subsequent events after the reporting date.
Specific audit procedures that have been impacted by the pandemic are physical inventory counts, asset inspection and testing, and performing walk-throughs of audit clients.5 These procedures have traditionally been undertaken in person. However, COVID-19 has made the execution of such procedures physically tricky, if not impossible. The use of remote collaboration technologies such as cloud computing, IoT, robotics and chatbots has become part of the discussion, as well as the use of machine learning (ML) in the development of decision aids that can assist auditors in performing their roles during these times of uncertainty.
The Digital Auditor
This new decade has set several records, such as the lowest oil prices produced in the United States,6, 7 the highest unemployment surge ever recorded in one week in the United States,8 and the “drop in business owners was the largest on record, and losses were felt across nearly all industries and even for incorporated businesses”9 as a result of the COVID-related restrictions. So, in this new situation, how can forecasts be accurately developed? How can trend analysis be performed when there is no comparison? There is a growing need for both audit clients and auditors to learn new economic, industry and client trends rapidly due to environmental and economic changes. A potential solution is presented in the power of ML, one of the subsets of AI.
ML is a technique that enables the analysis of massive data and detects hidden patterns, resulting in models that can perform prediction tasks. “Organizations have been quick to apply their AI and machine learning know-how in the fight to curb this pandemic.”10 A survey conducted by AvidXchange showed that “accountants have identified artificial intelligence and machine learning, real-time payments, and automation as the top three technologies that will impact the business-finance ecosystem the most this year.”11
Socially, as more of the human workforce shifts to remote work, several possible outcomes exist that are relevant to AI. Considering that one certified public accounting (CPA) organization listed digitization as one of its initial steps toward automation and the use of AI (figure 1),12 the move to working remotely has led to an automatic shift toward digitization. Less printing and scanning in the office and more typing and uploading digitally achieves a digitization step that enables more straightforward automation.
Once data is in a digitized form that is machine-readable, various AI techniques can be applied to it. In digital form, unstructured textual data can be analyzed by AI. For example, using Natural Language Processing (NLP) and machine vision (MV) (e.g., optical character readers), AI can analyze documents, such as thousands of invoices and contracts, within a few minutes. AI can then perform text clustering, sentiment analysis or text categorization. For example, to comply with the new lease regulation, the Big Four firms (Deloitte,13 EY,14 KPMG15 and PwC16) used AI to examine hundreds of thousands of clients’ leases instead of employing a manual approach, which would have been significantly less efficient.
IN DIGITAL FORM, UNSTRUCTURED TEXTUAL DATA CAN BE ANALYZED BY AI.
When the auditor is no longer physically present at the audit client site, what forms of AI can supplement the observation audit procedure? Not only do auditors obtain audit evidence through observation, but they are also able to pick up tangential information when physically present at their client’s premises, such as fraud cues. However, in a socially distant age, observation procedures are challenging. AI can augment the auditor in such a scenario. At the assurance level, the Big Four audit firms report using drones to perform inventory inspection at locations that are difficult to reach for auditors (e.g., oil rigs). For IT audits, the possibility of combining robotic process automation (RPA) with AI can automate the inspection of client system configurations. Instead of a point-in-time inspection, as is the case in many integrated financial statement audits, automated reviews using robotics can provide near-continuous auditing.
Digital auditing can work for the experienced auditor, but what about the less experienced auditor who relies on some form of mentorship or apprenticeship? How can AI bolster the performance of less experienced auditors? One angle to consider is the use of chatbots. One of the Big Four firms rolled out a chatbot for employees that answers general questions usually fielded by the help desk.17 According to the firm, the chatbot was able to answer 500,000 questions in 28 days. The return on investment for the chatbot was achieved within a week. Building a chatbot that can onboard new auditors is an efficient solution with minimal investment. All that is required is knowledge base of questions, which can easily be populated as new auditors interact with it. Experienced auditors can then provide answers through a live chat. Within a short time, the chatbot can learn from the responses of the experienced auditors and may be able to answer questions directly, with minimal live support.
AI Adoption and Audit Firm Size
The adoption of AI may result in a more significant productivity gap between small organizations and larger organizations, especially in times of economic uncertainty. Studies show that the Big Four firms are ahead of small and medium-sized audit firms regarding AI adoption and employee training on emerging technology.18 Smaller firms may not want to risk investing in sophisticated and risky technology such as AI in times of change. However, there is an option that smaller firms can choose, which may require significantly less investment in technology. AI-driven SaaS includes cloud-based platforms that organizations can implement based on the number of processes, users or other metrics.
One roadblock to AI SaaS is that audit firms may be unsure of cloud-based AI platforms in terms of data security and compliance with regulatory requirements. Other challenges include getting clients on board and obtaining the digital data required for AI use. Another huge problem is that the user firms may lack the technical know-how to rely on AI to perform some auditing roles and may be unable to articulate how AI performed its work in their audit.
THE ADOPTION OF AI MAY RESULT IN A MORE SIGNIFICANT PRODUCTIVITY GAP BETWEEN SMALL ORGANIZATIONS AND LARGER ORGANIZATIONS, ESPECIALLY IN TIMES OF ECONOMIC UNCERTAINTY.
However, in early 2020, organizations had to move to digital platforms to perform most business functions. Perhaps, in these times of uncertainty, organizations and regulators might become less resistant to cloud-based AI platforms and forge ahead with iterations of experimentation, validation and policy formation. SaaS platforms may provide small and medium-sized enterprises opportunities to explore the benefits of AI with minimal investment in hardware or IT skills. To be successful, there must be a strong collaboration between regulators, professional bodies, academia, AI developers and user firms (figure 2).19 An example of such partnerships within the auditing and accounting profession is the International Ethics Standards Board for Accountants (IESBA) Technology Initiative, which examined the ethical impact of AI within the profession.20 The initiative involved a collaboration of stakeholders, including global professional bodies, regulators, auditing/accounting firms, software vendors and academics. Another collaboration is the High-Level Expert Group (HLEG) on AI established by the European Commission, which developed the ethics guidelines for trustworthy AI. This expert group consisted of technologists, ethicists, educators, and industry and government representatives.21
Ethical Challenges With AI
When thinking of adopting AI, enterprises need to consider the ethical and operational implications of using the technology. Although the benefits of AI are immense, it is essential to consider the unintended consequences that may occur when using this technology. For example, when AI performs tasks such as the audit of system configurations, who will be accountable for issues that may spring from such use—the audit firm or the AI developer? Another ethical concern worth considering is what are the current limits of AI.
Although AI has been developed to handle specific tasks (i.e., narrow AI), there is yet to be an AI developed with full human capabilities, such as human emotional intelligence. Bearing this in mind, the profession may need to be continuously in touch with AI research advancements and challenges. A proactive approach to addressing the ethical challenges of AI is preferable to a reactive approach, as illustrated in figure 3.
Conclusion
In times of uncertainty, organizations need to be flexible enough to adapt to changes yet hold on to the core missions and principles that attracted their clients to them in the first place. By utilizing the opportunities that AI provides, bearing in mind the operational and ethical limitations of the technology, organizations can navigate these times of uncertainty with better decision aids and higher efficiency. AI implementation need not be a massive investment. Organizations can begin by exploring SaaS options and then build pilots for simple processes. Identifying the first AI use case is one of the most critical steps for any AI initiative. Plenty of AI platforms exist that require minimal coding to get a system up and running. With little required training, professionals can quickly implement pilot AI systems as a proof of concept before investing in the project. Auditing professionals interested in emerging technology can easily usher in a new era of innovation within the profession.
Endnotes
1 Liu, S.; “Artificial Intelligence Funding United States 2011-2019,” Statista, 2020, https://www.statista.com/statistics/672712/ai-funding-united-states/
2 Munoko, I.; H. Brown-Liburd; M. Vasarhelyi; “The Ethical Implications of Using Artificial Intelligence in Auditing,” Journal of Business Ethics, January 2020, https://link.springer.com/article/10.1007/s10551-019-04407-1
3 Bowling, S.; “How We Successfully Implemented AI in Audit,” Journal of Accountancy, 1 June 2019, https://www.journalofaccountancy.com/issues/2019/jun/artificial-intelligence-in-audit.html
4 Association of International Certified Professional Accountants (AICPA), “Audit and Financial Reporting Matters Related to COVID-19,” 2020, https://future.aicpa.org/resources/download/audit-matters-and-auditor-reporting-issues-related-to-covid-19
5 Mahbod, R.; M. Fredrickson; “Overcoming Site Visit Limitations in the Pandemic,” Journal of Accountancy, 8 May 2020, https://www.journalofaccountancy.com/news/2020/may/overcoming-auditing-site-visit-limitations-during-coronavirus-pandemic.html
6 Sönnichsen, N.; “Largest Slump in Crude Oil Prices During Coronavirus Pandemic By Type 2020,” Statista, 2020, https://www.statista.com/statistics/466293/lowest-crude-oil-prices-due-to-covid-19/
7 Kolakowski, M.; “History of Oil Prices,” April 2020, https://www.investopedia.com/history- of-oil-prices-4842834
8 Hansen, S.; “Biggest Unemployment Surge Ever: Record Weekly Claims Top 3.28 Million,” Forbes, March 2020, https://www.forbes.com/sites/sarahhansen/2020/03/26/weekly-unemployment-claims-surge-to-328-million-as-coronavirus-batters-the-economy/#7bd8b56511b0
9 Fairlie, R. W.; ”The Impact of COVID-19 on Small Business Owners: Evidence of Early-Stage Losses From the April 2020 Current Population Survey (No. w27309),” National Bureau of Economic Research, 2020, https://www.nber.org/papers/w27309.pdf
10 Sivasubramanian, S.; “How AI and Machine Learning Are Helping to Fight COVID-19,” World Economic Forum, 28 May 2020, https://www.weforum.org/agenda/2020/05/how-ai-and-machine-learning-are-helping-to-fight-covid-19/
11 Arrowsmith, R.; “Under Stress, Accountants Seek New Tech to Get Through the Pandemic,” Accounting Today, 9 July 2020, https://www.accountingtoday.com/news/under-stress-accountants-seek-new-tech-to-get-them-through-pandemic
12 Davenport T. H.; J. Raphael; “Creating a Cognitive Audit,” CFO, 12 July 2017, http://ww2.cfo.com/-auditing/2017/07/creating-cognitive-audit
13 Deloitte, 16 Artificial Intelligence Projects From Deloitte: Practical Cases of Applied AI, The Netherlands, 2018, https://www2.deloitte.com/content/dam/Deloitte/nl/Documents/innovatie/deloitte-nl-innovatie-artificial-intelligence-16-practical-cases.pdf
14 EY, “EY Lease Reviewer Global Proof of Concept Launches to Support Lease Accounting Changes,” United Kingdom, 2017, https://www.ey.com/en_gl/assurance/how-can-artificial-intelligence-enhance-your-lease-accounting-approach
15 Samuel, J.; “KPMG Applying IBM Artificial Intelligence to Help Businesses Efficiently Meet IFRS 16 Lease Accounting Requirements,” KPMG, 21 March 2018
16 PricewaterhouseCoopers (PwC), “PwC Collaborates With eBrevia to Deploy Machine Learning for Contract Analysis,” 4 May 2018, https://www.pwc.com/us/en/press-releases/2018/pwc-collaborates-with-ebrevia-to-deploy-machine-learning-for-contract-review.html
17 Orddmann, D.; A. Davies; “How Artificial Intelligence Can Enhance Your Lease Accounting Approach,” EY, 12 October 2017, https://www.ey.com/en_us/assurance/how-can-artificial-intelligence-enhance-your-lease-accounting-approach
18 Bakarich, K. M.; P. O’Brien; “The Robots Are Coming... But Aren’t Here Yet: The Use of Artificial Intelligence Technologies in the Public Accounting Profession,” Journal of Emerging Technologies in Accounting, 25 June 2020, https://doi.org/10.2308/JETA-19-11-20-47
19 Op cit Munoko, Brown-Liburd and Vasarhelyi
20 International Ethics Standards Board for Accountants (IESBA) Technology Working Group, IESBA Technology Initiative Phase 1 Final Report, USA, December 2019, https://www.ifac.org/system/files/meetings/files/Agenda-Item-5A-Technology-WG-Final-Report_0.pdf
21 High-Level Expert Group on Artificial Intelligence (AI HLEG), The Ethics Guidelines for Trustworthy Artificial Intelligence, European Commission, June 2018, https://ec.europa.eu/futurium/en/ai-alliance-consultation/guidelines
Ivy Munoko, CISA, ACCA
Is a Ph.D. candidate at Rutgers University (New Jersey, USA). She is also an online instructor, teaching a course on artificial intelligence (AI) for finance, accounting and auditing, where she provides hands-on tutorials on various AI techniques and their applications within the accounting and auditing professions. She has more than seven years of combined experience in IT, finance and auditing. She co-authored an article titled “The Ethical Implications of Using Artificial Intelligence in Auditing,” which will be published in the Journal of Business Ethics.