Journal Pre-proof

Implications of Artificial Intelligence on the Objectives of Auditing

Financial Statements and Ways to Achieve Them

Yubin Gao , Lirong Han

PII: S0141-9331(21)00208-8
DOI: https://doi.org/10.1016/j.micpro.2021.104036
Reference: MICPRO 104036

To appear in: Microprocessors and Microsystems

Received date: 2 December 2020

Revised date: 28 December 2020
Accepted date: 16 January 2021

Please cite this article as: Yubin Gao , Lirong Han , Implications of Artificial Intelligence on the Objec-
tives of Auditing Financial Statements and Ways to Achieve Them, Microprocessors and Microsystems
(2021), doi: https://doi.org/10.1016/j.micpro.2021.104036

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.

© 2021 Published by Elsevier B.V.

Implications of Artificial Intelligence on the Objectives
of Auditing Financial Statements and Ways to Achieve
Them
Yubin Gao *, Lirong Han 2
1

1. Business School, Beijing Technology and Business University, Beijing, China

2. Business School, Jilin University, Changchun, China
Corresponding Author’s Email: gaoyubin@btbu.edu.cn

Abstract
Artificial Intelligence (AI), thanks to its application in auditing financial statements, has exerted profound implications
on audit objectives and ways to accomplish them. In the past, however, confined by then-available technologies in the
profession, objectives of auditing financial statements were circumscribed within the conformity to the generally
accepted accounting principles, via accounting inspection as the only approach. With the advent of AI technologies,
objectives of auditing financial statements shall be positioned on the assurance of the reliability of accounting
information rather than the conformity of accounting reports to the fundamental principles of reporting preparation
standards. When striving for such audit objectives, auditors shall adopt more independent assurance methods other
than restrict themselves within the scope of accounting inspection. Audit propositions shall be derived from the
auditing purpose and objectives. Audit evidence shall come from more comprehensive and independent channels,
such as expert conclusions. By introducing the conclusions as a new kind of evidence, we may take the advantage of
AI technologies to build a deductive inference model that is scientific and comprehensive.
Keywords
Artificial intelligence, financial statements, audit theories
Introduction
Advancement and application of AI technologies are transforming the conventional life and work
patterns, resulting in indelible changes in the social environment. To better adapt the current
society where information and intelligence are swiftly evolving, all disciplines and professions are
restructuring or improving their strategies, organizations, products and procedures. The domain of
auditing is no exception. It now can employ electronic bookkeeping, data mining and multi-
dimensional data analysis. However, auditing technologies and procedures constitute merely a
subsection that AI is altering. It can exert resounding impact on the objectives of auditing,
especially on the objectives for auditors when audit financial statements and methods applied.
However as argued by Shi and Sun, in contrast to the rapid improvement of auditing practices
supported by information technologies, the framework of modern auditing theories is far from
being well-established, calling for extensive studies by more professionals. 1 Auditors shall, when
auditing financial statements, first specify the audit objectives and then break them down into audit
instructions to eventually reach the objectives. Audit instructions shall then chart the directions for
auditors to gather evidence. They then make their judgements, arrive at audit opinions and
complete the audit task. From establishing audit objectives to ways of achieving them, AI may
influence the whole procedure of auditing financial statements. This paper first analyzes how AI
may influence the objectives of auditing financial statements, and then continue with how
information technology influences ways of achieving audit objectives, including implications on
audit instructions, sources of audit evidence, formats of audit evidence and audit judgements.

How AI Influences the Objectives of Auditing Financial Statements

The objective of financial statement audit is determined by the audit purpose, which depends on the
demand of auditing. Different from other types of audit activities, financial statements auditing demand
from investors for the needs of the reliability of accounting information. Fair and reliable accounting
information can provide a basis for investors to make correct investment decisions, directing appropriate
allocation of economic resources, advancing capital market and the healthy development of the society as
a whole and its economy. However, for the sake of their own economic interests, enterprises may not
voluntarily provide fair and reliable accounting information in financial statements, which calls for the
need to establish audit standards. Abiding by the standards, auditors shall verify the financial statements
to provide reasonable assurance in terms of the fairness and reliability of the accounting information in the
financial statements. In this regard, the purpose of auditing financial statements is very clear, that is, to
assure the authenticity and reliability of the enterprises’ accounting information. As stated in Handbook of
International Standards on Auditing and Quality Control, the purpose of audit is designed to enhance the
degree of confidence of intended users of the financial statements. The purpose of auditing financial
statements is of service for the users of accounting information. To achieve such, auditing practitioners,
i.e. auditors shall act as practitioners of auditing financial statements and convert the audit purpose into
practicable auditing objectives. The conversion process may fall subject to the ceiling of the practitioner’s
capability and technologies available. When auditing standards regulate that auditing practitioners shall be
auditors, audit objectives are stated to fairly present an entity’s financial positions, results of operations
and cash flows. Fair presentation acts as the overriding factor amongst all audit objectives, through which
the credibility of financial statements could be attained. However, evaluation on fairness is restricted by
audit technologies. Manual auditing practices hardly provide sufficient support for auditors to arrive at
audit opinions on the fairness of the whole set of financial statements. Consequently, audit objectives,
during actual practices, are usually reduced to audit, in terms of all material matters, whether the
preparation of the financial statement is in accordance with the applicable financial reporting principles. If
financial statements are prepared in accordance with the applicable principles, then the presentation would
be deemed fair. This opinion has been recognized in audit theories, as reasoned by Mautz and Sharaf in
auditing postulates, that through consistent application of generally accepted principles of accounting,
financial positions and the results of operations would be fairly presented. 2 Current audit standards
resonate their opinions. As stated in Handbook of International Standards on Auditing and Quality
Control, the objectives of audit could be achieved, through the expression of auditors’ opinions, on
whether all material matters are reported in conformity to the applicable financial reporting principles. As
a result, the objective in practice of auditing financial statements has been shifted from the fairness of
financial information to the legitimacy of reporting procedure. It is a compromise, due to the actual
activities of manual auditing, that procedure legitimacy substitutes result fairness. While Mautz and Sharaf
regard the relationship between generally accepted accounting principles with fair presentation as one of
the audit postulates, they simultaneously point out that auditors only accept this assumption with more
reservation compared to any other assumptions. 2 Such objective is positioned with quite distance to the
audit purpose. Such a gap constitutes an imperative part of audit expectation gap, resulting in the
discrepancy between the profession of auditors with that of the legal service domain who represents the
interests of the general public.3
Though advancement in AI will not upset the demand to audit financial statements, rendering no change
to the purpose of auditing, it may bring significant changes in audit objectives. Considering the technical
support and safeguard measures AI can provide to assure the overall fairness of the entity’s financial
statements, audit objectives can accordingly be set with a clear positioning on the overall fairness of
accounting information, instead of the inference of fairness of the accounting information via the
legitimacy of the reporting process. As demonstrated in information economics, when some information is
derived from other underlying information, it is then referred as the second-order information, whereas the
fundamental information, which is rather crude, might be referred as the first-order information. Looking
at the principle, one may conclude that second-order information stems from the first-order information,
which, however, still possesses its idiosyncratic meaning as the independent information. What financial
statements presents is comprehensive second-order information, summarized from the first-order
information contained in accounting ledgers. When audited manually, second-order information may not
be directly verified, hence leading the auditing objective to infer the fairness of comprehensive
information from the legitimacy of the first-order information reporting.
AI technologies creates more access to a broader scope of data, along with fast and convenient
measures for crunching. The underlying data, after analysis, will generate a vast number of useful
information of multiple kinds that can directly verify the second-order information. In the process that is
in accordance with the time and space sequence inherent in the to-be-audited matters (e.g. trend, structure,
relationships and so on), auditors can set judgement rules and constraining conditions to establish
mathematical or logic formulas, which can be applied as technical measures to verify the actual time and
space status of auditing matters . 1 Deng pointed out that technologies like data mining could be applied to
enlarge those usually negligible differences to locate the imbedded drawbacks in the data structure. 4
stressing on the completeness and top-down auditing practices, Bell et al. used Business Measurement
Process (BMP) to establish appropriate expectation of the company’s accounting information and other
information and compare such expected value with the actual information in the accounting system of the
company to detect problems. 5
The ability to find abnormal accounting information in financial statements auditing is greatly enhanced
due to the availability of vast data and analysis accompanied by big dada mining models .We, as a result,
believe that with the increasing synergy between AI and audit practice, audit objectives should be really
positioned to ensure the reliability and fairness of accounting information, rather than the compliance of
accounting information preparation procedures, so as to reduce the gap of audit expectations and bridge
the gap between the accounting profession and the legal profession.
Influence of AI on Identifying Audit Instructions
Once audit objectives are set, it would then ensue the essential question on how to achieve them. To
accomplish such, one shall start with identifying audit instructions, around which gather and evaluate
audit evidence. Then auditors shall proceed with audit judgments, arrive at audit opinions and realize the
audit objectives. With more specifics, audit instructions bring audit objectives into sharper focus, forming
the prerequisite and foundation to collect audit evidence. According to the auditing theories, audit
evidence collection shall start with identifying audit instructions. Limited by audit technologies, assurance
on accounting information is usually conducted manually, which is highly dependent on inspecting
accounting ledgers, forcing auditors to opt for a quite realistic comprise to infer specified audit objectives
based on accounting assertions. Identification of audit instructions also become reliant on accounting
information assertions by the entity’s management. Enacted in 1980 by American Institute of Certified
Public Accountants, Amendment to Statement on Auditing Standards No. 31, “Evidential Matter” states
five assertions as representations by management: existence or occurrence, completeness, rights and
obligations, valuation or allocation, presentation and disclosure. These five assertions are explicit
embodiment of the recognition, measurement and reporting concerning accounting information. More
specifically, existence or occurrence, completeness, rights and obligations address the requirements on the
confirmation of accounting information. Valuation or allocation deal with the requirement of accounting
information measurements. Presentation and disclosure then address the requirements of accounting
information reporting. Meanwhile, Auditing Standards No. 31 also states that besides the auditing
objective of overall appropriateness, the other eight specific auditing objectives could be directly obtained
via above-mentioned five assertions. Such identification process of auditing objectives, not in view of the
audit purpose and the overall objective, is based on the confirmation, measurements and reporting of
accounting information. As a result, auditing practices would be reduced as accounting practice
inspection, rather than the assurance regarding the final output of accounting activities, i.e. accounting
information. To conclude, specific auditing objectives inferred from assertions are dislocated with the
overall auditing objective and is in serious deviation to the audit purpose, resulting in unclear and
undefined logic relationships amongst audit instructions, audit purposes and overall audit objectives.
Companies operate in an open, multi-dimensional and information-based environment, where
accounting information is affected. Consequently, the external environment provides the indispensable
foundation to build the auditing instruction system, in contrasting to the conventional instruction system
identified from accounting assertions. Auditing instructions, accordingly, could be open, multi-
dimensional and multi-layered, aiming at the auditing objective of the authenticity of accounting
information and effectively narrowing down the gap of auditing expectation. On the one hand,
restructuring the roadmap directing the identification of auditing instructions may take the reliance of
auditing practices on accounting inspection out of the equation. Instructions are identified considering the
auditing purpose, from which overall auditing objective is inferred. It may then be broken down into
specific audit objectives, which in turn give rise to audit instructions, assuring a rigid logic between
auditing purposes and auditing instructions. On the other hand, categories of auditing instructions could be
further expanded. On top of additional auditing instructions identified from accounting assertions during
the auditing phase, auditors could also identify instructions pertinent to auditing purposes. For example, to
assure the purpose of gathering authentic and reliable accounting information and its achievement,
auditing instructions that shall be identified includes, (1) overall appropriateness of accounting
information; (2) appropriateness of all specific accounting matter information; (3) appropriateness of
financial indicators in comparison to the sector peers; (4) appropriateness of financial indicators in the
perspective of historical trend of the company’s development; (5) appropriateness of the structure of all
financial data, etc. AI, supported by strong analysis function via big data searching, could not only satisfy
the requirements in the completeness of the set of auditing instructions, but also realizes the close
convergence between audit instruction system and audit objective system, pushing auditing practices
towards extreme proximity to the ultimate auditing purpose of the authenticity of accounting information.
Influence of AI on the Source of Audit Evidence
Auditing evidence comprises all factors that are influential on auditors’ thinking. As stated by Handbook
of International Standards on Auditing and Quality Control, No. 500, audit evidence refers to the
information used by the auditors to arrive at the conclusion, including relevant accounting information
contained in the financial statements and other information. This definition describes the source of
auditing evidence as being numerous; however, in actual practices, restricted by technology and cost,
auditing practices rely heavily on sampling. With limited information obtained and to be dealt with,
sampled evidences are usually finite in amount and unsystematic, showing weak proving power in terms
of the auditing instructions. Application of AI technologies enables auditors to supersede the scope of
business of the entity and sampled auditing. They can instead gather, analyze and process vast amount of
data from both the internal system and external environment of the listed companies, from historic records
to up-to-date realities, from accounting to non-accounting respects, to establish a three-dimensional
auditing evidence source from the space, history of development, internal structure of accounting
information of the companies. They can then obtain, analyze and process and process evidential matters
that corroborates each other and help auditors to judge true or false in terms of the auditing instructions.
Companies are not isolated islands in the market economy, whose existence and development are under
the implication of equity allocation, governance mechanism, management efficiency and production
efficiency; and at the same time, influenced by external sector development trend, macro-economic
environment, national policy and so on. As a result, there exists a high correlation amongst space, time
and internal structure regarding the accounting information of the companies’ financial positions, results
of operation and cash flows. The internal complexity extends far beyond the balances in double-entry
ledgers. This means that when auditors gather evidence, they also need to begin with these three
dimensions. First, competition evens the profit margin in any industry, especially for sectors with
complete competition, where hardly any company could keep long-term extra profit margin. Therefore, if
any specific level of a company’s accounting information deviates from the averaging trend of the sector’s
financial information, there would then speak for a high probability of irregularity in the dimension of
space in terms of the company’s on-going concern. In addition, company’s development usually follows a
roadmap, whose consistent progress follows a certain pattern, unless there being significant internal or
external transformations. The development of a company shall be progressive, forming a specific
development curve along the time sequence. Should there be a deviation of accounting information from
the development curve during any specific year, then the accounting information of the company is
encountered with an irregularity in the time dimension of company’s development. Lastly, considering the
internal logic and consistency of accounting data based on the double-entry ledger standards, should there
be information that is contradicting or hardly accounted for, then there might be irregularities in the
internal structural dimension of the accounting information. Using AI, auditors could inspect irregularities
and conduct deep mining on the three dimensions of the accounting information, to hunt for audit
evidence with a pre-defined purpose. From a macro-perspective, auditors then can arrive at an overall
assurance regarding the authenticity of accounting information. From a micro-perspective, they could
conduct deep mining on specific material risk matters. They can conduct detailed inspection on the whole
sample in terms of the key control points and assert the total from samples for non-significant risk. In this
way, auditors may effectively bring out the assurance function of auditing while enhancing its
independency.
Influence of AI on the Format of Auditing Evidence
Current procedures available for auditors to gather different kinds of information include inquiry,
inspection, observation, confirmation, recalculation, reperformance and analysis. Though stressed with
some level of importance in the auditing standards, group discussion and expert engagement may only
provide limited amount of audit evidence. If conducting manually, practitioners who are the external
personal of the entity, may hardly have access to substantial information. As a result, auditing mostly
depends on auditors and the auditing team. The advert of AI has generated opportunities of broader expert
engagement in the auditing domain, paving a way to explore more ways to gather auditing evidence.
Under such new circumstances, expert conclusions may be introduced as a new form of evidence for
financial statement auditing.
As a type of litigation evidence, expert conclusions are mainly applied in legal practices. As pointed out
by Wu, experts, commissioned or retained, utilize their professional know-how and modern science-based
technologies to inspect, analyze and make judgments, with the respect to certain specific matters arisen
from the litigation procedure, in order to issue written conclusions. 6 As a potentially new category of
auditing evidence, expert conclusions could also be applied in auditing practices. AI provides enormous
data information for external financial accounting professionals, finance and computing experts, who can
take the advantage of their own professional know-how to conduct in-depth analysis on the accounting
information via data mining and issue conclusive opinions on the overall quality of the accounting
information. Auditors may use expert conclusions as auditing evidence to proceed, not confined by
individual capacity in collecting evidential matters. This can safeguard the broadness and depth of
auditing practices exceed those of accounting practices, with enhanced auditing quality and ensured
authenticity and reliability of accounting information.
Influence of AI on Auditing Judgement
Under customary auditing model and due to the limitation of auditing theories and methods, after
gathering and assessing evidential matters, auditors then would need to rely on their individual know-how,
experience and wisdom to make judgements, draw auditing conclusions and eventually arrive at auditing
opinions. This explains why professionalism bears extra importance in auditing practices. However, one’s
cognitive capability is finite, so is the resources one could avail. When inferring, auditors may add their
subjective perspective, personal difference and thinking inconsistency, resulting in discrepancy between
audit conclusions and reality, or even incorrect auditing opinions. Comprehensive inference aided by AI, a
model with rationalism as the core, can effectively render practitioner’s judgement less subjective,
arbitrary and hard-to-verify.
AI improves the presence of auditing judgement independency in the auditing procedure. Shi and Sun
argued that for data-based auditing, the object to be audited is underlying electronic data on the books,
which means with information technologies available, auditors can establish economic information from
their independent system rather than the accounting information system, and utilize such economic
information to verify that in the accounting system. 1 As a matter of act, auditing practitioners have always
been exploring methods to independently verify the financial information generated from the accounting
entry system. Only such efforts have been largely bounded by manual auditing. During the early years,
gathering evidential matters solely relied on account sheets, which in nature was the inspection of
accounting matters, referred as general ledger accounting. However, to audit financial statement is to
assure the authenticity and reliability of accounting information, a purpose hardly achievable via simple
verification on the general ledgers. The falling of Mckesson & Robbins in 1938 justified the requirement
to obtain audit evidence through non-accounting matters. auditors shall first inspect the internal control
system and then test the account balances. This signals the start of the general system auditing, another
step of auditing assurance towards independence. To ensure the reliability of accounting information,
auditing enters risk-oriented phase, pushing the scope of evidence wider. Bell et al. proposed the auditing
evidence triangle theory, which states three sources of auditing evidence, one being entity business state
(EBS), another being management information intermediaries (MII), and the other being management
business representations (MBR).7 Above opinions states that the assurance procedure shall be independent
from accounting balance system. Whereas without AI’s support, auditors may be presented with
mountainous difficulties when endeavoring to gather evidence outside of the accounting system and
internal control system, resulting in the limited capacity of manual work to obtain and process auditing
evidence. Application of AI can restructure auditing process, liberating it from the constrains of
accounting system. Relevant economic information of the company generated from more independent
channels gives rise to a new form of auditing evidence, which independently verifies reliability of
accounting information for sake of the ultimate audit purpose. As a result, the concept of auditing
independency shall be expanded under the availability of information technology, that auditing procedure
shall stand independently from accounting procedures, assuring the objectiveness and effectiveness of
auditing practices.
Conclusions
Regarding the content, the purpose and objectives of auditing are aligned, only presenting themselves
through different formats from various perspectives. From the standpoint of an investor, auditing practices
are motivated by the need to prevent financial fraud and protect their economic interests, inducing the
purpose of auditing to assure the authenticity and reliability of accounting information. For auditors, it is
necessary to define the audit objectives according to the audit purposes, then decompose the audit
objectives into specific audit propositions, collect audit evidence around the audit propositions, draw audit
conclusions through audit judgments, and finally achieve the audit objectives. Since auditing objectives
are formulated by auditors in accordance with auditing purposes, they are equivalent to each other in
nature. However, when auditing is conducted manually, its instructions, source and format of evidences
and judgements are all under the constraint of auditing technologies, knowingly comprised and weakened
during the auditing procedure. Thus, audit objectives are deviated from the purpose, reducing the level of
confidence that investors, creditors and other stakeholders have in the audit reports prepared by the
auditors. AI, developing quite fast, has exerted significant influence on data processing, auditing
environment, sources and formats of evidences, audit practitioners and matters as well as audit
judgements, reforming the measures and tools of certified accountant to audit financial statements and
backing intelligent audit with technology application. From the respect of auditing system, Liu
summarized a seven-step procedure of computer-based auditing.8 Yang and Huang proposed an admissive
intelligent audit model.9,10 AI has rendered qualitative changes in audit procedures, establishing an open,
multi-dimensional, multi-layered system for audit instructions. In terms of the source of audit evidence, it
helps to establish a three-dimensional source of space, time and the internal structure of accounting
information. In terms of the format of audit evidences, it introduces the possibility of applying expert
conclusions as a new format. Regarding audit judgements, it develops a comprehensive inference model
with the core being rationalism. AI technologies converges the audit objectives of the auditors with those
of information demanders’, narrowing the gap of audit expectation and effectively improving the
credibility and market position of auditors (As shown in Figure 1).
 demander implementer

investor auditor

confirm
execute
audit motives

AI technologies converging the audit objectives and audit purpose audit audit
proposition evidence

decompose
audit audit audit process
audit purpose
assignment objective
audit audit
audit objectives opinion judgment
audit objectives deviated
from the purpose deviated from
the purpose the preparation of
the authenticity and reliability faithful presentation of financial statements on
of accounting information financial statements the basis of framework
for financial reporting

the traditional auditing technique leading to the deviation of the audit objective and the audit purpose

Figure 1. Theoretical framework of the relationship between audit objectives and audit objectives

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, author- ship, and/or publication of
this article.

Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.

Biography

Gao Yubin , Male, was born in June 1987 in Ye County, Henan Province, China, Ph.D. in management
(accounting)of Jilin University, Postdoctoral researcher in applied economics (finance) of the China Center for
Industrial Security Research(CCISR) at Beijing Jiaotong University. In May 2018, he entered the accounting
department of the Business School of Beijing Technology and Business University, and he devoted himself to
the capital market research in accounting and auditing issues. He has published more than 10 papers on various
publications, for example the Auditing Research, Contemporary Economic Research, Contemporary
Accounting Review, and participated in the preparation of a graduate student's teaching material. He has partly
edited the graduate teaching material, and participated in many national or provincial research projects.
HAN Lirong was born in Meihekou, Jilin Province in July 1963. She is a professor and PhD supervisor in
the School of Business in Jilin University where she was awarded with a PhD degree in Economics. She is
dedicated to researches in capital market accounting, auditing issues and studies in auditing systems
regarding certified accountants. Prof. Han has published over 30 dissertations on Auditing Research,
Contemporary Economic Research and Contemporary Accounting Review. She has led multiple
researches funded by the state and provincial ministries.

Yubin Gao

Lirong Han
References
1. Shi A Z and Sun J. Preliminary Interpretation on Audit Models. Auditing Research 2005;4: 3-6.
2. Mautz R K, Sharaf H A. The Philosophy of Auditing. Sarasota: American Accounting Association1961:36-37.
3. Liu Y. Where is the disagreement between the field of law and that of accounting? China Certified Accountant
Correspondence1998; 7: 28-38.
4. Deng Q S. Study on Identifying Information Distortion of Listed Companies Based on data mining. Doctoral dissertation,
Jiangxi University of Finance and Economic, Nanchang 2009:48.
5. Bell T B, Marrs F O, Solomon I, Thomas H. Auditing Organizations through a Strategic-System Lens. [EB/OL] 1997;
http://business.illinois.edu/kpmg-uiuccases/monograph.pdf.2014.5.15.
6. Wu Z Y. Evidence Law. Zhengzhou University Press, Zhengzhou 2005:79.
7. Bell T B, Peecher M E, Solomon I. The 21st Century Public Company Audit: Conceptuals Elements of KPMG's Global
Audit Methodology. KPMG International 2005:9
8. Liu R Z. A few matters on current computer-based auditing. Auditing Research (Supplement) 2006: 3-6.
9. Yang J and Huang CY. A study on inducting and intelligent audit mode. Journal of Shanghai Business School 2008; 3: 90-
94.
10. Yang J and Huang CY. A further study of the import intelligent audit. Journal of Shanghai Business School 2009;3: 95-99.