The Possibility of Artificial Intelligence Adoption in Heading Human Auditors into Extinction
By: Jessie Danielputri Yuwono 2101704546
CHAPTER I INTRODUCTION
Objective
The purpose of this research is to discuss the current and the upcoming technology development that has impacted and will further impact how an auditor conducts his works. Specifically, is to analyze whether artificial intelligence (AI) will bring more benefits or a threat instead for the profession, which is the likelihood to lead the existence of human auditors towards extinction.
Background
For such a long time, audit profession has been well-known for the hard challenges and tough mentality needed. During the interim year, for example, it might put the auditors at the expense of being home at 3 AM yet have to be ready again at 8AM. Auditors are never allowed to be wrong and thus are always put at a high risk of being misleading. The consideration received for bearing those conditions and other implicit benefits have consequently triggered many young generations to become ones. However, it will take different requirements compared to it was, owing to the significant changes as a result of advanced technology.
Artificial intelligence (AI), the rapidly evolving area of technology, has come as the ‘game changer’ for various industries. AI has proven to be beneficial in helping human daily activities. Although so, the existence of AI has also brought a concern such as the replacement of current existing jobs by computers. Oxford University (Frey, C. B., 2018) through their recent study estimated that amongst 600 professions being considered, auditor was at the top list of being deemed ripe for automation with 96% chance to be replaced in two or three decades to come. In fact, Herbert Simon (1960) had previously predicted on his essay that computers held the comparative advantages in rule-based activities that would lead to decline of routine jobs (i.e. inspecting documents) but never mentioned a timeline for how long it takes before such jobs would disappear.
As mentioned by Cathy Engelbert (cited in Cohn, M., 2016), Deloitte’s CEO, the auditing profession might be going to change more in the next five years than it has in the latest 30 years. The ambiguity of whether AI will head to the replacement of human auditors or bring value-added for auditors would be discussed in detail on later sections.
CHAPTER II LITERATURE REVIEW
2.1. Auditor’s Role in Present Technology
The main objective of conducting an audit is to provide reasonable assurance that information reported on the client’s financial statements contains no material misstatement and agrees with the established criteria, such as IFRS as the related accounting standard. Generally speaking, is to improve the quality of financial statements and/or to reduce the information risk so that users would not put themselves on business risk decision.
To achieve that, auditors must collect appropriate and sufficient supported evidences, evaluate, and verify again. While doing so, auditors, combined with their manual-conducted audit activities, make use of the evolving technology in various ways. The example of the most basic adaptation of technology in conducting audit is the usage of Argus tools by Deloitte in their audit engagement, in which the firm can transfers their potential clients’ transactions data to the analytics team abroad, receive the results’ graphics in return, and thus the firm can identify certain points of audit to stress on before finally put into further analyzing process. As claimed by Yulia The (Jon, R., 2015) as the innovation team leader of Deloitte Southeast Asia, the mentioned tool is believed to increase the efficiency of an audit and reduce up to 30% of human error possibilities.
One of the biggest accounting fraud scandal, HealthSouth, in which the auditor was ‘tricked’ by the management, could not be detected for such a long time beforehand due to the traditional sampling method used that often contains auditors bias (Daigle et al., 2013). Fortunately, the presence of today’s ongoing technology allows data analysis on complex transactions, hence, results in proper target of the audit and can help in detecting such fraud scheme earlier.
More advanced technology in auditing such as auditing software that provides more accuracy on their works, and to alter the time efficiency as well will be explained in details on section 2.2.
While auditors have received the positive impacts in making use of the technology (i.e. software), their roles are not separated from the current success of such technology. To illustrate, audit software as a basic machine learning model can only function according to the program that the auditor has initially designed. However, if the algorithm returns into an inaccurate result, the auditor must then intervene and make adjustment over and over again, until the machine learns by its own along the time.
2.2. AI and Its Involvement in Auditing
Entering the industry revolution 4.0, the complexities resulted from undergoing big data and connected world could only be managed by the phenomenal advanced technology, the artificial intelligence (AI). AI is portrayed to be the activity devoted to make machines work as intelligent as human brain is, often represented by computer systems, in performing human-like tasks appropriately (Nilsson, N. J., 2010). However, are AI products necessarily smarter than the origin human brains? The answer is relied on the credit one is willing to give to the synthesized software and hardware. To illustrate, electronic calculator computes numbers faster than that of manually done by human, and hardly ever make a mistake. Standing on the perspective that such intelligence relies on multi-dimensional spectrum (Grosz et al., 2016), the difference between arithmetic calculator and human brain is not on the kind, but on the scale, speed, degree of autonomy, and generality. Such argument is also applied to evaluate other instances of AI, such as SIRI, speech recognition software, to self-driving cars. Notwithstanding the sophisticated yet popular picture it has, however, the precise definition of AI is still difficult to pinpoint.
According to Piletsky (2019), the development scenarios of AI includes that of a machine on the level of being unconscious to that of being conscious. The former represents the weak AI or what is called as artificial narrow intelligence, whereas the latter represents the strong AI or known as artificial general intelligence.
Narrow AI system is programmed to perform a single specific task, operates only the pre-determined and pre-defined range, and thus is unable to perform any task outside the scope it is designed for. The machine learning discussed on earlier segment is within this phase of AI. The real proof of a machine learning involvement in audit profession is the usage of robotic process automation (RPA) system such as software, namely ACL or even the expensive one, Continuous audit software, which is beneficial for being able to, for example, finding unbalance and duplicated journal entries. The software can process data way faster than auditors would manually do, and thus relieve the auditors from devastating realities in doing monotone tasks. Auditors also thus no longer need to perform infinite tracing and vouching activities. Additionally, by leveraging AI, auditors can assess and study larger data, even up to 100% complete population in contrary with the traditional audit method at which auditors are limited to samples only. However, again, it has no self-consciousness to think for itself.
This narrow AI system acts as the building blocks of the coming of more intelligent AI—the deep learning as the subfield of machine learning—on the upcoming general AI phase. At general AI, machines are expected to be able to do problem solving, put judgments under uncertainty, plan, learn, integrate prior knowledge in decision making, or in other words, to perform intellectual task that requires human cognitive ability.
The common question being wandered is, at what states of AI are we in at the moment? A venture capitalist and a former AI researcher, Kai-Fu Lee (2018) defined the current pace as “age of implementation” at which the technology is just spilled out of the lab and begin to enter the world. The statement is thus a step backward in AI time line from when AI is going to be normalized as Benedict Evans (2018), the fellow AI researcher, stated “We will be there at point of everything will have machine learning somewhere inside and no one will care.”.
The applications of deep learning that will encounter auditors’ works in near future are already reflected briefly at the moment. To illustrate, the Big 4 firms’ examples will be used. Ernst & Young (EY) has recently applied deep technology to scan an accounting lease contract to identify the relevant clauses for accounting treatment like the commencement date, amounts, renewal or termination date. As analyze subsequently by EY Global Assurance Innovation leader, Jeanne Boillet (2017), the AI tools can electronically review up to 80% of a simple lease contents and around 40% of complex leases with notes that the figure for the latter is likely to increase as the machine keeps on learning. This technology is then beneficial in a way that the administrative time needed to review audit documents can be reduced and thus, allow the auditors to spend more time on more crucial areas such as risk identification and business insight. Furthermore, the number of contracts than can be reviewed by such AI tools are way larger and in much shorter timeframe. Another deep technology that EY just launched is the use of airborne drones to monitor inventory during performing audit procedures, which are physical count and observation. For instance, this drone can count the number of vehicles in a client’s plant and observe the quality of related items to determine the presence of obsolesce, if any. Nevertheless, the two of EY’s AI tools mentioned above seem to be in an R&D mode at the time being and have not been applied broadly in auditing their clients.
CHAPTER III DISCUSSIONS
“Why spend precious limited time counting how many individual fish have expired, when you can focus on finding out what’s killing the fish in the first place?”
3.1. The Future of Audit
On his book, Dan Zitting (2017), ACL’s chief product officer, shared that if he could turn back to his time as auditor, he would have used different audit approach other than the traditional one that makes the whole audit seems to be performed just because it is mandated to. Instead, he would also bring up the context that the client is actually concerned about, which is a risk-based discussion like “based on key risks around revenue assurance, we see indicators that $ xxx of billings are at risk due to weak customer engagement.” Such performance can be accomplished by the presence of AI in audit that allows auditors to have more time examining the more complex and higher risk area that require judgment for high level of uncertainty.
He (ibid.) further breaks down the next generation of audit techniques and technologies into:
- Data analytics
- Mobile devices as core audit tools
- Continuous audit
- Real time, automated, assurance-related reporting and dashboard
With those technologies be altogether, audit in years to come is predicted have the capability in providing better level of assurance than the reasonable assurance in audit of today (PCAOB, 2017).
In the future, auditors will also have the ability to access client data in a timelier manner and on standardized format, or what is known to be continuous audit. The benefit brought will be the ability to reduce financial loss occurred from risks of fraud that is used to caught post the event, because auditors are going to monitor transactions continuously (Viet, T., 2019). This reflects the analogy stated in italic sentence above that future audit will not be about spending time finding any possible concealed frauds, but instead preventing one from happening.
Furthermore, as the AI cognitive-based tools permit auditors to access a wide-range of clients’ information, the assurance will be expanded beyond just the financial statement, but also include cybersecurity and sustainability reporting (Yew, S. W. et al., 2015). Hence, indicates the changing role of an audit committee than it is nowadays.
After all, the increase of automation due to AI adapted might head to the future where CPA firms will likely hire fewer junior auditors who previously performed the manual tasks that at that moment, will have been automated thoroughly. Predicted by Steve Varley (cited on Grut, O., 2016), the chairman of EY UK & Ireland, the fresh graduate recruitment in accounting and auditing work areas could drop to 50% by 2020 while EY itself as part of Big 4 and the industries are one of the biggest employers in UK, even in the world. CPA firms are also predicted to compete with technology companies namely Apple or Google in terms of designing algorithms.
3.2. The Impact AI Gives to Audit Profession
While it is most likely that cognitive technologies will augment large portions of both public and private companies’ audits in the future, a stigma born within the communities that the audit profession will come to a fully-automated audit with zero percent of human auditors involvement on the horizon.
It is indeed true that cognitive technologies will replace the rote tasks that have been conducted manually for many years, such as doing a physical count for inventories. However, it does not necessarily mean the full elimination of human auditors. Instead, such automation gives pro-active value-added for auditors.
Nevertheless, it is also true that the work field in the industry will not be as big as it probably is at the moment. Thus, the competition amongst the job seekers, especially the fresh graduates, would be very intense, including those who are now already established as auditors. Many of traditional auditors are coming toward obsolescence and do not even realize it for they are rarely focus on things that are really matter. As a result, they will be systematically replaced by automation as the big data technology is advancing.
Therefore, whether the existence of AI tends to be a threat to replace auditors or no is highly depended on how one can justify with the significant changes resulted. There is a clear difference between the auditors who just sit silently and watch the technology safe and sound, and the auditors who are proactively developing automation and drive value and efficiency from that. AI will not replace auditors, but auditors that interpret AI will replace those who do not. Therefore, the ones being automated will be replaced by natural selection while the one creating the automation will survive despite the rapid competition.
The new transformed audit techniques in the future will link to new required skills that future auditors must master. A solid fundamental in accounting is and will forever be the basic essential, but as the auditors’ role becomes more insightful, a broader enhanced skills are needed. That is to be knowledgeable and capable in science, technology, engineering, and mathematics (STEM) as well as strong capability to work with large number of data and strong analytical skills.
Nonetheless, no single university with accounting program nor post-hire training programs has developed a full curriculum with a complete set of advanced audit analytics such as statistics, cognitive technologies, data management, and other necessary subjects (Davenport, T. H., 2019). BINUS International, for instance, is working on altering the formula for the best curriculum that will be suitable for this audit transformation.
References
Billet, J. (2017). Ersnt & Young: How AI will enable us to work smarter, faster. Retrieved
from https://www.ey.com/en_gl/assurance/how-ai-will-enable-us-to-work-smarter-faster
Cohn, M. (2016). Deloitte Leverages AI and Analytics for Audit. Retrieved from
https://www.accountingtoday.com
Daigle, Ronald, J., Louwers, T.J., & Morris, J. T. (2013). HealthSouth, Inc.: An Instructional
Case Examining Auditor’s Legal Liability. Issues in Accounting Education, 4(93).
Davenport, T. H., Raphael, J. (2019). Creating a Cognitive Audit. CFO Publishing LLC.
Retrieved from https://www2.deloitte.com/content/dam/Deloitte/us/Documents/audit/us-audit-creating-cognitive-audit.pdf
Evans, B. (2018). Ways to think about machine learning. Retrieved from
https://www.ben-evans.com/benedictevans/2018/06/22/ways-to-think-about-machine-learning-8nefy
Frey, C. B., & Osborne, M. (2018). Automation and the future of work—understanding the
numbers. Retrieved from https://www.oxfordmartin.ox.ac.uk/opinion/view/404
Grosz, B. J., Altman, R., Horvitz, E., Mack worth, A., & Mitchell, T. (2016). Artificial
Intelligence and Life in 2030. One Hundred Year Study of Artificial Intelligence.
Grut, O. (2016). AI could destroy hiring in one of the biggest industries for graduates. Retrieved
Jon, R. (2015). Deloitte wins ‘Audit Innovation of the Year’ at 2015 International Accounting
Bulletin awards. Deloitte Press Releases. Retrieved from https://www2.deloitte.com/us/en/pages/about-deloitte/articles/press-releases/deloitte-wins-audit-innovation-of-the-year.html.
Lee, Kai-Fu. (2018, October 22nd). The Four Waves of A.I. The Fortune. Retrieved from
http://fortune.com/2018/10/22/artificial-intelligence-ai-deep-learning-kai-fu-lee/
Nilsson, N. J. (2010). The Quest for Artificial Intelligence: A History of Ideas and
Achievements. Cambridge, UK: Cambridge University Press.
PCAOB. (2017). Technology and the audit of today and tomorrow. Retrieved from
https://pcaobus.org/News/Speech/Pages/Harris-statement-PCAOB-AAA-4-20-17.aspx#_edn5
Piletsky, E. (2019). Consciousness and Unconsciousness of Artificial Intelligence. Future
Human Image, 11, p.69.
Vasarhelyi, Miklos, A., Warren Jr., Donald, J., Teeter, Ryan, A., & Titiera, R. R. (2014).
Embracing The Automated Audit. Journal of Accountancy.
Viet, T. (2019). PWC: Continuous audit & monitoring. Retrieved from
https://www.pwc.com/vn/en/services/consulting/continuous-audit-monitoring.html
Yew, S. W., Houw, T. I., Gan, T. T., Lim, D., & Leong, K. (2015). Cybersecurity: The changing
role of audit committee and internal audit. Deloitte. Retrieved from https://www2.deloitte.com/content/dam/Deloitte/sg/Documents/risk/sea-risk-cyber-security-changing-role-in-audit-noexp.pdf