See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/351946460

Impact of Artificial Intelligence on Auditing -An Exploratory Study

Conference Paper · September 2020

CITATIONS READS
8 6,338

2 authors, including:

Ravi Seethamraju
The University of Sydney
68 PUBLICATIONS 1,384 CITATIONS

SEE PROFILE

All content following this page was uploaded by Ravi Seethamraju on 27 October 2022.

The user has requested enhancement of the downloaded file.

Association for Information Systems
AIS Electronic Library (AISeL)

AMCIS 2020 Proceedings Accounting Information Systems (SIGASYS)

Aug 10th, 12:00 AM

Impact of Artificial Intelligence on Auditing - An Exploratory Study

Ravi Chandra Seethamraju
University of Sydney, [email protected]

Angela Hecimovic
University of Sydney, [email protected]

Follow this and additional works at: https://aisel.aisnet.org/amcis2020

Recommended Citation
Seethamraju, Ravi Chandra and Hecimovic, Angela, "Impact of Artificial Intelligence on Auditing - An
Exploratory Study" (2020). AMCIS 2020 Proceedings. 8.
https://aisel.aisnet.org/amcis2020/accounting_info_systems/accounting_info_systems/8

This material is brought to you by the Americas Conference on Information Systems (AMCIS) at AIS Electronic
Library (AISeL). It has been accepted for inclusion in AMCIS 2020 Proceedings by an authorized administrator of
AIS Electronic Library (AISeL). For more information, please contact [email protected].
 Impact of Artificial Intelligence on Auditing

Impact of Artificial Intelligence on Auditing –

An Exploratory Study
Completed Research

Ravi Seethamraju Angela Hecimovic

University of Sydney Business School University of Sydney Business School
[email protected] [email protected]

Abstract
Artificial Intelligence (AI) assisted by the massive processing power and big data, though is poised to
influence the auditing profession, empirical evidence is sketchy. This exploratory cross-sectional qualitative
field study, using the Technology-Organization-Environment (TOE) framework as its broad theoretical
lens, and semi-structured interviews as the data collection strategy, investigates the adoption of Artificial
Intelligence (AI) in auditing and identifies the factors influencing the adoption. Adoption of AI by auditing
firms, our study found, would not only result in improvement in audit quality and decision making but also
helps them build capabilities to deliver value-adding AI consultancy services that are more profitable than
traditional auditing. Apart from improving audit quality, challenges auditing firms face include
conservative approach of external regulatory bodies, inability to document the use of technology for
verification by regulators and perceived legal and reputational risk of financial auditing work that limits the
potential use of AI.

Keywords

Auditing, Artificial Intelligence (AI), TOE framework

Introduction
The Advances in emerging technologies such as data analytics, blockchain and artificial intelligence (AI)
are expected to influence the accounting and auditing. Though the deployment of AI in auditing is not new,
its impact is expected to be more substantial now because of the availability of massive processing power
and big data (Kokina and Davenport 2017). Lack of productivity improvement in the past decade and the
increasing sophistication of the software and hardware that can perform various types of cognitive tasks,
are accelerating its use. For accounting firms, the labor-intensive nature of the traditional auditing process
and the increasing demands for compliance with laws, regulations and policies, have made it imperative to
use such emerging technologies for productivity improvements (KPMG 2018).
Several initiatives are being trialed across the world and the Big Four accounting firms EY, Deloitte, KPMG
and PwC are investing millions of dollars in AI with the objective of building capabilities to deliver more
cost efficient and high-quality audits to clients. For example, KPMG has started employing IBM Watson’s
deep learning systems to analyze banks’ credit files for commercial mortgage loan portfolios, while Deloitte
in collaboration with Kiva systems employs AI to review contracts, leases and invoices (Sun and Vasarhelyi
2017). Similarly, PWC uses Halo, a machine learning tool to analyze journal entries and identify problem
areas (Dickey et al. 2019). With a prediction that 30 percent of the corporate audits will be performed using
AI by 2025 (World Economic Forum 2015), the field of auditing is in for a significant change (Agnew 2016).
Australian business organizations are reportedly slow to adopt technologies and AI is no exception.
According to a report published by Accenture Australia (2019), Australian businesses are well behind their

Americas Conference on Information Systems 1

 Impact of Artificial Intelligence on Auditing

US and European counterparts in adopting state-of-the-art AI, and the productivity growth also lags US
and Europe (AFR 2019).
Research to date into the adoption of AI on auditing is limited and most of the publications were descriptive
in nature, highlighting the ‘potential’ of AI. Whilst advances in Artificial Intelligence (AI) are projected to
fundamentally change the field of auditing, very little empirical evidence is available about its antecedent
factors and adoption challenges. Other than a discussion of the emerging technologies, potential benefits
and challenges and research agenda, empirical research is limited. For example, Sutton et al (2016) and
Issa et al (2017) on the applications of AI and its potential detrimental effects and technology dominance,
Kokina and Davenport (2017) on role of transparency and bias in AI-based decision making in auditing,
and Duan et al (2019) on AI technology-human interaction and AI implementation.
Our study aims to fill this gap and provide empirical evidence on the factors influencing the adoption of AI
in auditing. Based on semi-structured interviews with senior auditors, regulators and AI experts that are
amid observing, shaping and influencing the use of AI technology in Australian audit practice, our study
aims to investigate its impact on audit process and identify the factors influencing the adoption. By
analyzing the potential and actual use of emerging AI technologies on the audit and across the various stages
of the audit process, this study aims to uncover and address pertinent issues in incorporating AI into the
audits. By adopting Technology-Organization-Environment (TOE) framework as a broad theoretical lens,
this study aims to identify antecedents and challenges in the adoption and use of AI in audit process. The
findings of the proposed study would be of relevance to regulators and standard setters and facilitate the
creation of guiding principles and frameworks for the adoption and use of AI in financial audits. Though
anecdotal, our study found increasing use of artificial intelligence in auditing in Australia.
This paper is organized as follows. It will first review the literature on AI in auditing and highlight the
potential benefits and challenges reported in the literature. It will then briefly explain the Technology-
Organization-Environment (TOE) framework, a theoretical anchor used, and the cross-sectional field study
methodology employed in the data collection and analysis in this study. Next sections present findings and
discussion and conclusions.

Literature Review
Broadly defined, AI is a computing technology that exhibits some form of human intelligence and covers
several interlinked technologies including data mining, machine learning, natural language processing,
speech recognition, image recognition and sentiment analysis (EY 2018). The potential power of AI lies in
its ability to recognize patterns and make predictions that facilitate decision-making (Goh et al. 2019).
Auditing is a process that has not changed much for years (AICPA 2012). AI displayed the ability to disrupt
industries that typically consist of repetitive and predictable tasks (Chui et al., 2016). As the audit process
typically involves recurring, high volume anticipated transactions, AI has the potential to impact the audit
process (Baldwin et al. 2006). With capabilities to efficiently analyze large quantities of data, AI would not
only enable auditing of 100% of financial statement data but can accelerate auditors’ work (Issa et al. 2016;
Bizarro and Dorian 2017). Adoption of AI, it is argued, could improve judgement on audits and auditor’s
decision making (Sun and Vasarhelyi 2017). In fact, it is argued, such AI-based judgements are more
efficient than humans (Wu, et al. 2015). Traditional manual audit procedures are inefficient as humans are
considered less proficient in tasks that require collecting and analyzing a large volume of transactional data
(Dai and Vasarhelyi 2016; Issa et al. 2016). Therefore, it is argued that could be useful in audit procedures
such as materiality and risk evaluation, control assessment, audit planning, opinion selection and reporting
(Vasarhelyi and Kogan 2017; Bierstaker et al. 2014). Other benefits touted in the literature include
reduction in human errors (Murphy 2017), facilitation of continuous auditing (Brennan et al. 2017), and
capacity to audit all transactions and lower costs and auditing time (Issa et al. 2016; Westhausen 2016).
Adoption of technologies in auditing is not new. Auditing has successfully adopted various information and
communication technology tools such as audit toolkits, logit models and control templates into their
operations to achieve greater efficacy levels (Manson et al. 2001). Dowling and Leech (2014) have observed
use of broadly referred computer-assisted audit tools and techniques (CAATs) that include intelligent
decision aids, expert systems and audit support systems. Uses of CAATs for data testing, integrated test
facility, parallel simulation, embedded audit module and generalized audit software are well documented
in literature (Bierstaker et al. 2014).

Americas Conference on Information Systems 2

 Impact of Artificial Intelligence on Auditing

Auditing firms in US and Europe have started deploying different strategies for building AI capabilities to
help them develop niche specialization and to gain competitive advantage as stated in their websites
(Omoteso 2012; KPMG 2018; Ting and Vasarhelyi 2018). For example, KPMG was working with IBM
Watson to develop “Supervised Learning”, a system where auditors assume the role of teaching AI
procedures in executing logical and rational thinking alongside using professional judgement and
skepticism where AI learns and develops itself to handle tasks. KPMG also is testing an AI solution that
helps identifying data in credit files, extracting usable data, and recognizing trends and patterns (KPMG
2018). PricewaterhouseCoopers (PwC) is developing an AI tool “GL.AI” that identifies abnormalities in
clients’ general ledgers by examining data from over one billion journals in over 4 countries and is expected
to analyze billions of values in seconds to identify abnormal transactions (PwC 2020).
Deloitte uses a natural language processing technology that retrieves information and data from
populations or samples and then transfers to a platform for automating the reviewing process for
contractual information, thereby cutting the time spent on document reviewing by half (Deloitte 2019).
Similarly, another firm, EY uses machine learning to review and examine lease contracts with an accuracy
level of 97% and uses robotic process automation in confirmations allowing auditors to use their
professional judgement and skepticism to determine the confirmation’s validity (EY 2018). EY Australia
claimed that half of their audit confirmations they requested for was completed using this AI and has future
plans to implement drones to monitor clients’ inventory during the audit (EY 2018).
All the examples of adoption discussed above are by big four accounting firms and on a trial basis. Some of
the factors that are limiting the adoption include, high initial and maintenance costs of AI systems, limited
knowledge of technology-enabled auditing among graduate auditors, legal aspects of reliance on AI-based
systems in audit (Issa et al. 2016). Further, clients may prefer interaction with auditors rather than with
machines, which highlights the continued importance of conversations and rapport building (Chan et al.
2012). Inaccuracies in the input data, potential human bias and incomplete data in the AI solutions
(Brynjolfsson and Mcafee 2017; McCollum 2017) may lead to the view that AI is still a “black box” that is
difficult for humans, even experts to understand (McCollum 2017). Even with the influx of Big Data, data
quality can be a big problem leading to a “garbage in garbage out” scenario where poor-quality data leads
to analysis results that are unusable (PwC 2020). Though Anderson et al. (2003) found some anecdotal
evidence to indicate that auditors preferred interactions with intelligent machines over client interactions,
concerns over data security, AI integrity, and lack of guidance on the relevant auditing standards are viewed
as constraints to their adoption (Issa et al. 2016).
At organizational resources level, the skillsets required to work with these emerging technologies are
different. AI has already started influencing the hiring practices in the Big Four firms where they prefer
individuals with fundamental accounting knowledge along with programming skills (Cooper et al. 2019), in
addition to data management skills (Ovaska-Few 2017). Adoption of AI, it is argued, could enhance
auditors’ technical skills including coding and visualization to understand and manipulate massive volumes
of data from both internal and external sources (Deloitte 2018; PwC 2020). Using AI will help achieve this
objective through data-driven decision making, data analytics to derive actionable insights and freeing up
accountants to work on value-adding tasks rather than being swamped by tedious grunt work (Goh et al.
2019). Marr (2018) infers that AI is unable to mimic some of the unique capabilities of auditors such as
judgement, emotional intelligence and professional skepticism. Brennan et al (2017) argue that the role of
auditors could change with the addition of responsibility to interpret data produced by AI and transferring
it to their clients to add value to their businesses.
Thus, there are several issues that require attention by researchers for the accounting firms to exploit the
AI technologies better and drive economic benefits. As pointed out by Banham (2018), most of the anecdotal
evidence of AI use in practice is in US context, and the applications are mostly confined to automation of
some routine tasks or data analytics solutions. Empirical evidence in Australian context is limited. This
study therefore aims to fill this gap in research and investigates the impact of AI adoption on auditing in
Australia. Next section explains the theoretical framework.

Theoretical Framework
Technology adoption though plays a complementary role in generating sustained competitive advantage,
the reasons for adoption and the nature of their impact varies from one organization to another and one

Americas Conference on Information Systems 3

 Impact of Artificial Intelligence on Auditing

profession to another. This is contingent upon varying contexts within which these technology adoptions
are situated including external competitive factors, regulatory and professional bodies and internal
organizational factors such as processes, management support, level of technology sophistication and the
technology itself. Though there are several theories that deal with technology adoption, the Technology-
Organization-Environment (TOE) framework developed by Tornatzky and Fleischer (1990) is considered
appropriate to analyze the adoption of AI in auditing. The TOE framework posits that technological,
organizational and environmental factors influence the adoption and adaptation of technological
innovations by firms and individual professionals. It integrates contingent technological, organizational
and environmental factors faced by auditors and auditing firms with the evaluation, implementation and
use of AI during different stages of audit process. Using this framework, this study aims to understand the
adoption of AI in the audit process including the influence of external, organizational and technology
related factors.
The proposed methodology is qualitative, cross-sectional field study, where the data was collected primarily
from semi-structured interviews (Lillis and Mundy 2005). This type of field research is appropriate given
that the use of contemporary AI tools within the audit profession is relatively new and our interpreted new
insights (Ahrens and Chapman 2006) may shed light on how auditors (and other stakeholders such as
educators, standard setters are adapting and shaping their practice to this emerging technology. Given the
exploratory nature of the research, interviews based on individual perceptions and perspectives of the key
individuals in the auditing firms were undertaken. For data collection, a loosely structured interview guide
allowing flexibility was employed to conduct the interviews. Respondents were asked about the general
perception of AI, potential benefits and challenges of adopting AI technology on auditing, and the
organizational, environmental and technological factors that may influence their effective adoption were
examined. After obtaining necessary ethics clearance, so far 14 interviews were completed. It consisted of
seven audit managers/partners (respondents A1, A2, A3, A4, A5, A6 & A7), two standard setters (A8 & A9),
three AI and auditing software experts (A10, A11 & A12) and two auditing/assurance experts (A13 & A14).
All the respondents have considerable experience in the field. These research interviews were recorded with
prior permission and transcribed for further analysis. Earlier, two pilot interviews were conducted in order
to test the process of questioning and structure (Yin 2009) and the interview guide was refined. A semi-
structured interview protocol and recording the data mechanically using a digital recorder were employed
to improve reliability. Data validation was carried out firstly by recording the interviews and transcribing
them verbatim, and by sending them to individual respondents for validation and corrections, and finally a
chain of evidence was established using verbatim transcripts and notes of observations made during the
interview. The data thus collected is coded and analyzed with reference to the themes identified from the
theoretical framework and literature.

Analysis and Findings

The analysis starts with the perceived impact of AI on various stages of auditing. It will then discuss the
barriers and challenges to the adoption of AI considering TOE framework as the basis. Insights generated
from this study, despite the risk of generalization, can deliver some pointers and offer foundations for
further research.

Impact on audit stages

AI could influence different stages of auditing. Audit planning, the first phase, is considered an iterative
process and is critical first step in ensuring that auditors focus on the areas of greater risk. As noted by a
senior audit partner, “AI will hit traditional audit firms specifically in the audit process and function as
there will be a lot of mechanical function that will be more efficiently performed by AI and machine
learning techniques” (A3). Another respondent opined that the audit stages are no longer clear cut. “We
need to rethink the way we organise the audit” (A11). It is claimed that “none of the firms is really using
any AI on their external audit engagements yet and I don’t see that coming up very soon” (A11). Further,
it is claimed that “firms and audit clients are reluctant (to adopt AI) as data access is challenging” (A10).
Despite this reluctance to change and adopt, there are some auditors that would welcome AI to assist in the
planning stage. The following quote clarifies this: “Like we have a lot of independent steps in planning that
could be easily done with AI…like review your board minutes and tell you what is interesting? Like tell
you what is appearing like maybe risky parts of the business. So yeah in the planning stage. And there

Americas Conference on Information Systems 4

 Impact of Artificial Intelligence on Auditing

are so many things that it could help us with that. Like in Client X, they have like 300 individual legal
entities and we have to form the point of view of related parties to keep like a database of all entities that
we have an assurance deliverable for. So as part of independence we have to reconcile the organisation’s
structure to what we have in the database to make sure we have included every entity to make sure we
have covered everything. So it’s like a mapping and it takes SO long. And Client X ends up buying and
investing every month so we do this task every month and it takes 2 girls to do this offshore a week to do
it properly… AI could help” (A1).
The risk assessment aspect in planning is a key judgment area that AI has the potential to “extracting
outliers, say how we [auditors] should treat this and if there are risks” (A4), and in assisting to “make
critical judgement on risky areas of client’s operations” (A7), and in “summarizing key risks during the
year you should be looking at” (A6). While AI could assist in risk identification, it cannot possibly analyse
the risk implications. Auditors would still be needed to determine if anomalies being pulled out are actual
risks utilising professional judgement, our study observed, as AI is found to take significantly longer to
replace tasks requiring judgemental skills (Agrawal et al., 2017).
Prior literature (Goh, 2019; Dai and Vasarhelyi, 2016) has pointed out the role played by AI through
extensive use of algorithms and models, to assist in gathering evidence to detect fraud in client’s operations.
Our study found AI to be useful in the evidence gathering phase of the audit process. As observed by a senior
auditor, “AI reduce testing time, testing the whole population; the biggest issue at the moment is sampling,
testing in the field and confirmations” (A1), and “especially the lower level use of AI in confirmation
processing and analysis” (A6).
There is also potential to impact the reporting stage. As pointed out by a respondent, AI could identify risk
and give an indication of the potential implications for the report. “Feeding through the data of the evidence
collecting stage, using the whole population really and pulling out outliers and saying how we should
treat this and if this is at risk or not, giving me the profile, to show what is of greater risk. Then all the
way to the end, when it comes to reporting, it could probably tell me what the potential implications be
for the audit report” (A6).
Importantly, it is believed that the audit stages traditionally known would either disappear or merge with
others with the adoption of AI. As noted by an auditor, “thinking about audit stages is almost obsolete,
those stages will merge, because if we are looking at the whole population [with AI potentially], we are
going to identify where are the risk areas [planning stage], isn’t that then providing you with evidence
and so I think the test of controls and substantive testing [evidence gathering stage], the boundaries might
eventually merge, because we might be doing the same thing, where the processes, we might have the 3
way matching. We will have the substantive testing and the test of controls” (A4). Another observed that
“I think, the definition and ideas of the principle we are living with in audit, and separate stages and the
different types of testing it is all going to merge, because these AI tools would allow us to do that” (A4).
Regarding opinion/reporting phase, Fisher et al (2016) argued that natural language processing could be
used to help identify a company’s risk by interpreting data from sources such as social media. Respondent
A4 agreed that a “more interesting use of AI is risk evaluation and interpretation of evidence”, suggesting
that AI could augment audit quality and judgement by providing more evidence and reduce bias in the
process. As noted by a senior auditor, “doing manual stuff like gathering evidence, allows auditors more
time to have professional judgement, and (contribute to) audit quality increases” (A7), and that “audit
quality would improve due to higher assurance from entire population testing” (A6). Another noted that
“you can have this technology where we recommend you to test 48 transactions for this level of materiality
of this transaction, which are random to test the riskiness of this population…so you get a truly
independent selection and in theory, gives you a better outcome” (A9).
Another observed, “…when you got machine that actually does it fast, and doing all the stuff that otherwise
a human being would be digging around, you’re then using the cognitive powers of a human being to then
apply judgement. Which I think if auditors started to audit companies, based on the decision-making
drivers, what drives the decisions, that leads to this sort process, or that kind of reasoning, I think what I
think would be the next stage that AI, could get auditors to become. And once you get to that stage, I think
being an auditor would become much more satisfying and defined (A5).
The audit industry is starting to see a disruption as these Big Four accounting firms are offering AI
consulting services and working on joint development of specific AI tools as these services would potentially

Americas Conference on Information Systems 5

 Impact of Artificial Intelligence on Auditing

be more profitable than audit. A large-scaled, high-priced audit conducted by traditional auditing firms
could potentially be less attractive when compared with other firms providing such services using AI.

Technology related factors

As discussed in the literature review, the idea of employing some form of artificial intelligence tools in
auditing is not new and in the past tools such as intelligent agents, audit support systems and expert systems
were used for selective tasks. AI technology is multifaceted and has several components and is perceived
differently by different people. Its impact on auditing is also dependent upon these perceptions and varies
significantly depending upon individual professional’s experience and expertise. While some respondents
termed it as an umbrella term for data analytics and machine learning (A7) and that it is abstract (A9),
others believed that AI can do the thinking for humans (A1, A5) and has the ability to replicate how humans
think (A6). Confirming this ambiguity, our study found differences in the perceptions of respondents as to
what AI encompasses and therefore its potential impact on auditing.
Reflecting a simplistic view of AI, some of the respondents in our study, suggested that AI is like using
CAATs and that AI can assist in developing patterns and judgements while examining large volumes of
homogenous transactions with similar characteristics. AI simply provides more support and information
and critical judgement is always human. What seems apparent is that AI is considered just another tool,
and whilst AI is perceived as useful in terms of delivering potential efficiency to the audit, there is limited
evidence in relation to that yet from audit practice. As pointed by one, “this (AI decision) does not make
sense based on this data, or these transactions are usual… it will give you more support and information
rather than the small samples” (A4) and that “you will still need the auditors for their critical judgement”
(A10). Being a technology that is still evolving, having a diverse view seems appropriate. Further, another
noted that “AI, in my mind there are different levels, there can be automation, or a model that you can
build and add logic and data in and you analyze it and make it predictive or like there could be machine
learning where it could be unstructured or structured where you can tell the machines what to do or they
can find out what to do. But I think all of those are AI” (A3).

Environment related factors

External entities such as audit clients, auditing standard setting bodies, professional organizations and
regulatory bodies, have significant influence on the adoption of AI. Their conservative nature, and lack of
adequate technology skills and competencies within those organizations could limit the adoption and use
of AI technologies, our study found. Though AI has powerful capabilities and ability to explore grey areas
which were hitherto resolved using human judgment, its adoption, is heavily dependent upon the external
bodies such as standard setters and regulatory bodies such as ASIC (Australian Securities and Investment
Commission). Respondents expressed concerns as to whether regulators, standard setters and audit clients
are ready for AI in Australia and that clients may not be equipped to handle exceptions from reporting. As
pointed out by a respondent, “client acceptance is the main issue where the tools on offer are perceived to
be immature” (A2). Other clients, the respondent A3 noted, have expressed concerns with regard to privacy
and security of the information. Other challenges are documenting the use of technology and technology-
savviness or lack of it by the regulatory bodies. As noted by another, “the biggest issue is not the use of the
technology, but how do we document the use of this technology. Like how you write it in an audit file, like
how do we use it in this. Because when ASIC (Australian Security and Investments Commission) comes,
they’ll be like I don’t understand that (A1) and…ASIC isn’t exactly very technologically upskilled. And even
the audit standards, like ASA 315 or 520, none of them actually have got records of firms using bots or
AI…like look at how long it took them to come up with the standard like 20 years or something... like how
long are they going to get them to come on board with this AI…(A1). A representative from the Australian
standard points out the challenges in the regulatory space given that the AI solution is a ‘black-box’ to many.
“People are worried about using AI in this regulatory space [like ASIC] where it’s like does AI meet the
standards? (A2).
Professional organizations are another external entity that need to focus on upskilling auditors by
incorporating technology content into their professional examinations and certifications. Interviewees
agree that content on technology such as data analytics and IT should be added into syllabus as “it is
apparent that current University accounting and auditing degree don’t cover the data and analytics side
of the auditing and computers side enough” (A9).

Americas Conference on Information Systems 6

 Impact of Artificial Intelligence on Auditing

Organizational factors
It is argued that adopting AI technology may contribute to potential for cost savings through reduction of
jobs and in the process may replace people with valuable skills. Flagging the impact, a respondent (A10)
pointed out the potential loss of knowledge workers’ roles in auditing firms and the consequent loss of
valuable auditing and corporate knowledge. Unlike other technologies, AI, if implemented effectively, could
perform the tasks typically performed by several auditors efficiently and reduce the cost of service to the
auditing firms and thereby the costs to the clients. As noted by a respondent, “AI is past the point that it is
replacing cashiers, or jobs that don’t need a degree. It is starting to replace people who have valuable
skills for the last decade. A lot of people are unhappy with AI, or scared, which I understand, but there are
a lot of people who are excited, because it can solve their business problems (A10).
There also seems to be a capacity building exercise in professional firms that needs to occur to facilitate AI
adoption. Interviewees currently undertaking audits, A1, A2, A3, A5 and A6 have each identified their firms’
approach and investment into emerging technologies and that practice is varied. While some firms are
setting up a separate labs/units to develop specific solution for a specific task/problem, test them and give
them to the audit teams as required, other firms are establishing a multi-disciplinary teams that comprises
of experts in AI and data analytics along with traditional auditors. As pointed out by a respondent, “we now
have this thing called the digital lab and it is an app store of sorts and it changes the way we [auditors]
think from ground up instead of top down. And cause like so many of our grads are from computer science
background now and they have so much of their own ideas…you can always put it in as a request at the
lab and they can build a digital app… the idea is where you sit in an audit team and identify technology
problems and build codes and stuff to quickly do something (A1). Another noted that “…in terms of AI, we
have a push towards robots and data analytics right now, so we can reduce a lot of physical work efforts
from auditors themselves also make everything more efficient and the clients are really buying it in terms
of this approach … is actually intriguing from our clients right now...they invested a lot of money and
there are tools being developed (A3). Thus, big-4 accounting firms, as expected, are looking towards their
global arms to test and deliver those solutions for them to use and achieve productivity improvements.
Thus, cost savings and ability to deliver quality audits with the help of AI technology are key drivers for
accounting firms.
Though regulators and standard setters are often blamed for not clearly articulating the AI elements in
standards and processes, internal organizational factors such as quality data, good access and appropriate
skillsets are more important for the AI initiative to be successful. Client organizations are required to invest
in those resources to derive best advantage. A partner in auditing firm, observes the following: “A lot of the
practitioners are blaming standards and the regulators that they can’t really do it [use AI in audits],
because the standards not clear and they can’t comply with the standards. But I think the reality is that it
is quite costly, it requires a lot of time, and a lot of expertise. You need to have infrastructure in place in
clients, where you can get access to data, and data that is useful and then, the skillsets”.
Auditors’ general reluctance to trust the machines for judgement and therefore their belief that AI
technology cannot correctly identify the unusual transactions, is a factor that limits the adoption. The
following quote exemplifies this: “because partners take full ownership of a project and how can you take
ownership of everything the AI is producing if you don’t know how it is being produced. Like if you can’t
look at the codes, like most of the partners I don’t think can code and can’t understand” (A2)
Quality of data used as an input to the AI program is another factor. Bias can be injected into the AI program
if the data quality is poor and inaccurate, and which may in turn influence the human judgment. This is
perceived as a huge impediment to using AI in auditing, our study noted. As pointed out by a senior audit
partner, before AI can be adopted, “one of the biggest challenges is getting good data” and “clients are slow
to invest in digitization and data quality and still have piles of paper on their desks”. (A1). Therefore,
“there is a lot of bias in their data sets. I’m not saying they’re stupid. I think AI is such a new field and
people are discovering this problem overtime and I think removing bias in data sets is something that is
really important. So that will be one of the challenges. That will require thinking more of the data sets but
also checking for bias and also checking in your outcome (A4)
Another important area that was evident from the interview data is the required element of trust in AI,
considering it as a black box. Such hesitance is expected considering the legal implications of financial
auditing work and need for the auditors to confidently express an opinion and report based on the data

Americas Conference on Information Systems 7

 Impact of Artificial Intelligence on Auditing

produced by the AI system. The following statement by a senior audit partner illustrates this point. “With
AI, you got to trust it for it to work, like analyse something and tell you like predictive and tells you trends
like. I think it’s a bit more intense and more people would be more hesitant towards it. Cause it is an
unknown. And I suppose it learns the more you do it. Like an unknown fear… that we as auditors need to
overcome…(A1).
Resourcing with relevant skill sets is an important requirement. With emphasis on the future auditor to be
technologically-savvy, organizations are investing in training and upskilling their auditors. Data suggests
that being technologically-savvy does not necessarily mean learning how to code or write a language
software, like R or Python. It is more about understanding the data output and what the technology is
capable of. As suggested by one partner, it is “not exactly about knowing the language tool, but knowing
what the tool can do, like drive decisions” (A3).

Conclusions
The role of AI in auditing is increasing rapidly with major implications for the quality and process auditing.
Our qualitative exploratory study provides anecdotal evidence on the impact of AI on auditing and auditing
stages. Though it is in infancy stage of adoption in Australia, its impact is well recognised, and efforts are
on way, to prepare the firms (especially the big 4) and profession to take advantage of this technology and
help in improving audit quality and process. In addition to the investment in these technologies, some firms
are setting up separate labs/units to develop AI based solutions while others are establishing multi-
disciplinary teams that comprises of experts in AI and data analytics along with traditional auditors. With
such capabilities and adoption of AI on auditing process, audit firms could offer not only new value-adding
consulting services that are more profitable than traditional auditing, but also improve audit quality and
decision making. As noted by a senior partner, big auditing firms will “continue with their advertised
investment in AI, as they don’t want to be on the backfoot but a lot of them are using AI in consulting, tax
and looking through R&D, expenses and trying to identify what is deductible but until people feel
comfortable with what is coming out of the black box, they won’t move into audit…. yet” (A1).
External entities such as audit clients, auditing standard setting bodies, professional organizations and
regulatory bodies, it appears, are constraining the adoption and use of AI technologies, due to their
conservative approach to technology adoption and lack of technology-skills. Though they are often blamed
for not clearly articulating the AI elements in standards and processes, quality data and good access with
the client and appropriate skillsets in the auditing firm are critical for the AI initiatives to be successful, our
study found. Though, AI has powerful capabilities to explore grey areas which were hitherto resolved using
human judgment, audit clients in Australia, are concerned with privacy and security of data, have limited
ability to handle exceptions generated by AI applications and inability to document the use of technology
for regulators. Human bias built in the AI program because of the poor quality of input data and lack of
trust on AI solutions because of the ‘black-box’ are impediments to the adoption, study observed.
Considering the legal and reputational implications of financial auditing work and the need for auditors to
confidently express an opinion based on the data produced by the AI system, such hesitance in the adoption
can be understood. Many stages of auditing will be influenced by the adoption AI and Audit stages
traditionally known would either disappear or merge with others with the adoption.
Though our study makes an important contribution to the literature by exploring the technological,
environmental and organizational factors influencing the adoption of AI in auditing, it has two limitations.
First its data collection is limited to Australian context and therefore has limited generalizability in terms
of its findings. Secondly, the use of cross-sectional field study, while methodologically better than a single
case study, suffers from the limitation of drawing definitive conclusions that are embedded in Australian
context. Further in-depth case studies to understand how AI is changing auditing practice and auditing
profession will offer deeper insights and further refinement of theories.
AI experts and technology partners of the leading auditing firms can use the factors identified in our study
to improve the solution development and consider improving customer support to expand AI use by
auditing firms. Findings of our study will help auditing firms and auditors to understand the stages of
auditing process at which the adoption could be most effective. Given the significant investments required
for the technology solution and building skills and capabilities, understanding the potential impact of AI
use on the audit judgement, audit quality and performance would help firms to prioritize their resources.

Americas Conference on Information Systems 8

 Impact of Artificial Intelligence on Auditing

Given the complexities associated with the use of AI, auditors face increased processing and coding
demands and further emphasis on critical thinking and professional skepticism. The future auditor will be
expected to possess sound accounting knowledge underscored by a strong proficiency in accounting
technologies and AI capabilities and be skilled in communication to be able to interpret and communicate
the data effectively to their management and audit clients.

REFERENCES
Accenture Australia. 2019. Accenture Technology Vision 2019, Melbourne: Accenture Australia.
Agnew H., 2016. “Auditing: Pitch battle,” Financial times (May 9),
(https://www.ft.com/content/268637f6-15c8-11e6-9d98-00386e39d accessed 19 October 2019).
Agrawal, A., Gans, J., & Goldfarb, A. 2017. “What to Expect From Artificial Intelligence,” MIT Sloan
Management Review (58:3), pp. 23–27.
Ahrens T., and Chapman C. 2006. “Doing qualitative field research in management accounting: Positioning
data to contribute to theory,” Accounting, Organizations and Society (31), pp. 819–841.
Australian Financial Review (AFR). 2019. Why Australia is falling behind in productivity, 20 June 2019,
(https://www.afr.com/technology/wages-hurt-by-low-tech-adoption-and-less-job-switching-
20190619-p51zbc accessed 20 November 2019).
Baldwin, A., Brown, C., & Trinkle, B. 2006. “Opportunities for artificial intelligence development in the
accounting domain: The case for auditing.” Intelligent System in Accounting, Finance and
Management, (14:3), pp. 77-86.
Bierstaker, J., Janvrin, D., and Lowe, D. 2014. “What factors influence auditors’ use of computer-assisted
audit techniques?” Advances in Accounting (30:1), pp. 67–74.
Bizarro P. A. and Dorian M. 2017. “Artificial Intelligence: the future of auditing,” Internal Auditing (32:5),
pp. 21-26.
Brennan, B., Baccala M., and Flynn, M. 2017. “How Auditing Will incorporate AI,” CFO Auditing,
(https://www.cfo.com/auditing/2017/02/artificial-intelligence-audits/ accessed 20 November 2019).
Brynjolfsson, E. and Mcafee, A. 2017. “The business of Artificial Intelligence, what it can and cannot do for
your organization,” Harvard Business Review (http://alertlogic.qa.hbr.org/cover-
story/2017/06/the-business-of-artificial-intelligence accessed 20 November 2019).
Chan, H., DeBoskey, D. G., and Hee, K. 2012. “Audit fee patterns of Big Four and Non-Big Four firms,” The
CPA Journal, (82:10), (https://www.questia.com/magazine/1P3-2826779561/audit-fee-patterns-of-
big-four-and-non-big-four-firms accessed 20 November 2019).
Chui, M., Manyika, J., and Miremadi, M. 2016. “Where machines could replace humans-and where they
can't (yet),” McKinsey Quarterly, July (https://www.mckinsey.com/business-functions/digital-
mckinsey/our-insights/where-machines-could-replace-humans-and-where-they-cant-yet accessed
19 January 2020).
Cooper. L., Holderness. D. K., Sorensen. T., and Wood. D. A. 2019. “Robotic Process Automation in Public
Accounting,” Accounting Horizons, (33:4), pp. 15-35.
Deloitte. 2019. Using AI to unleash the power of unstructured government data, Deloitte Insights
(https://www2.deloitte.com/insights/us/en/focus/cognitive-technologies/natural-language-
processing-examples-in-government-data.html accessed 19 January 2020).
Dowling, C. and Leech, S. 2014. “A Big 4 Firm’s Use of Information Technology to Control the Audit Process:
How an Audit Support System is Changing Auditor Behavior,” Contemporary Accounting Research
(31:1), pp. 230-252.
EY. 2018. How artificial intelligence will transform the audit
(https://www.ey.com/en_gl/assurance/how-artificial-intelligence-will-transform-the-audit accessed
5 February 2020).
Goh, C., Pan, G., Seow, P .S., Lee, B.H.Z., and Young M. 2019. “Charting the future of accountancy with AI,”
Research Collection School of Accountancy (https:/ink.library.smu.edu.sg/soa_research/1806).
Fisher, I., Garnsey, M., & Hughes, M. 2016. “Natural Language Processing in Accounting, Auditing and
Finance: A Synthesis of the Literature with a Roadmap for Future Research: NLP in Accounting,
Auditing and Finance,” Intelligent Systems in Accounting, Finance and Management (23:3), pp. 157–
214.

Americas Conference on Information Systems 9

 Impact of Artificial Intelligence on Auditing

Issa, H., Sun, T., and Vasarhelyi M., A. 2016. “Research ideas for artificial intelligence in auditing: The
formalization of audit and workforce supplementation,” Journal of Emerging Technologies in
Accounting (13:2), pp. 1-20.
Kokina, J. and Davenport, T.H. 2017. “The Emergence of Artificial Intelligence: How Automation is
Changing Auditing,” Journal of Emerging Technologies in Accounting (14:1), pp. 115-122.
KPMG. 2018. “AUDIT 2025: The future is now,” Forbes Insights.
(https://i.forbesimg.com/forbesinsights/kpmg_audit2025/KPMG_Audit_2025.pdf accessed 19
October 2019).
Lillis, A.M. and Mundy, J. 2005. “Cross-sectional field studies in management accounting research –
closing the gaps between surveys and case studies,” Journal of Management Accounting Research
(17), pp. 119-141.
Manson, S., McCartney, S., and Sherer, M. 2001. “Audit automation as control within audit firms,”
Accounting, Auditing & Accountability Journal (14:1), pp. 109-30.
Marr, B. 2018. “How much data do we create every day? The mind-blowing stats everyone should read,”
Forbes, (https://www.forbes.com/sites/bernardmarr/2018/05/21/how-much-data-do-we-create-
every-day-the-mind-blowing-stats-everyone-should-read/#6317e1b560ba accessed 5 February
2020).
McCollum, T. 2017. “Audit in an age of intelligent machines: Already in use at many organizations, artificial
intelligence is poised to transform the way business operates,” Internal Auditor (74:6), pp. 24-29
(https://iaonline.theiia.org/2017/Pages/Audit-in-an-Age-of-Intelligent-Machines.aspx accessed 5
February 2020).
Murphy, H. 2017. “Auditing less of a burden as accountants embrace AI,” Financial Times.
(https://www.ft.com/content/0898ce46-8d6a-11e7-a352-e46f43c5825d accessed 5 February 2020).
Omoteso, K. 2012. “The Application of Artificial Intelligence in Auditing: Looking back to the Future,”
Experts Systems with Applications: An International Journal (39), pp. 8490 – 8495.
Ovaska-Few, S. 2017. “How artificial intelligence is changing accounting,” Journal of Accountancy,
October, (https://www.journalofaccountancy.com/newsletters/2017/oct/artificial-intelligence-
changing-accounting.html accessed 5 February 2020).
PwC. 2020. What is an audit? Available at: https://www.pwc.com.au/assurance/audit.html, (accessed 5
February 2020).
Sun T., and Vasarhelyi M., A. 2017. “Deep Learning and the Future of Auditing: How an Evolving
Technology Could Transform Analysis and Improve Judgment,” CPA Journal (87:6): pp. 24-29.
Ting, S. and Vasarhelyi, M. A. 2018. “Embracing Textual data analytics in Auditing with Deep Learning,”
International Journal of Digital Accounting research (18), pp. 49-67.
Tornatzky, L. and Fleischer, M. 1990. The Process of Technology Innovation, Lexington: Lexington Books.
Vasarhelyi, M.A. and Kogan, A. 2017. “Artificial Intelligence in Accounting and Auditing. Towards a new
paradigm,” Rutgers (4), (http://raw.rutgers.edu/MiklosVasarhelyi/
Resume%20Articles/BOOKS/B13.%20artificial%20intelligence.pdf accessed 5 February 2020).
Westhausen, H. 2016. “The tech-savvy auditor: Effective use of audit technology can enable audit
departments to provide valuable insights,” Internal Auditor (73:3), pp.18-21.
(https://iaonline.theiia.org/2016/Pages/The-Tech-Savvy-Auditor.aspx accessed 5 February 2020).
World Economic Forum 2015. Deep Shift: Technology Tipping Points and Social Impact. Available at:
http://www3.weforum.org/docs/ WEF_GAC15_Technological_ Tipping_ Points_ report_2015.pdf,
(accessed 5 February 2020).
Wu, Y., Kosinski, M., and Stillwell, D. 2015. “Computer-based personality judgments are more accurate
than those made by humans,” In Proceedings of National Academy of Sciences of United States of
America, (112:4), pp. 1036-40.

Americas Conference on Information Systems 10

View publication stats