Running head: ETHICAL IMPLICATIONS OF AI IN AUDITING

The Ethical Implications of Artificial Intelligence (AI) in Auditing Business

Name [By]

Affiliation of Institution

Name of Professor

Course

Date
ETHICAL IMPLICATIONS OF AI IN AUDITING 2

Introduction

Pew Research Center (2019) records that Artificial-oriented intelligence (AI) applications

are fast becoming more available in the current society as it is recorded that more than 81% of

Americans are phone owners. For instance, a mere search of an article in the google engine

search will show more advertisements on various social media or websites. AI recommendation

engines initiate the process involved in these multiple advertisements. The recommendation

engines create a custom list for the user through machine learning. It is AI in filtering

information about the preferences of the user. This is just but part of many examples of how AI

can be used in solving problems that seem difficult to solve while using the traditional

approaches. The use of AI technology is currently implemented in many business sectors, and

there are expectations that it will impact the auditing business. Therefore, understanding how

crucial the AI functions are (Kokina & Davenport, 2017).

There is a rapid evolvement of AI technology. It is paramount that people in the auditing

business understand the AI capabilities and their impacts on the decision-making process in

auditing. Additionally, the ethical impacts related to the use of AI are issues that also affect the

auditing business. Given its practical use and burgeoning development of data analyses and

interpretation, AI is essential to the auditing business. As recorded by Baldwin, Brown, and

Trinkle (2007), emerging and new tools are necessary for the effectiveness and efficiency of

tasks. Many firms that are providing auditing services search for ways of using AI to assess the

risks of semi-structured and unstructured repetitive decisions. The use of AI is still in its early

development stages in the auditing business, although its application in various aspects of

modern society is prominent. The general objective of this study is to become more enlightened

on the implications of AI in the auditing business.

Artificial Intelligence Patterns

Walch (2019) states that seven patterns are used in combinations of different projects that

involve AI. The first pattern of AI is hyper-personalization, which uses machine learning to

develop a unique profile of individuals. Over time, these profiles adapt to be used in providing

content that is personalized and unique to each other, rather than grouping people into categories.

Hyper personalization is applicable in targeted advertising in creating individual users' profiles

and displaying content and products relevant to the previous preferences of the users. The second

pattern of AI is Autonomous systems that can complete goals and tasks and interact with their

environment with little or no human involvement. The autonomous system can possess both

software and hardware components, though the ultimate goal is to minimize human labor.

Autonomous systems are applicable in boats, airplanes, and cars. The third pattern of AI

is predictive analytics and decision support. Walch (2019) notes that this process uses

machine learning and cognitive approaches to determine patterns that help predict future

outcomes. Predictive analysis is also used in projection methods like forecasting, which help

make better decisions by humans.

The fourth pattern of AI is conversation and human interaction, which is a vital part of

AI because there is still a need for human interaction. However, there is technological

advancement in Artificial Intelligence. The main aim of this pattern is to enable the interaction of

machines with humans in similar manner interactions are done between humans (The Seven

Patterns of AI, 2019). Machines can communicate with humans through audio, images, text

generation, chatbots, and voice assistants. The fifth AI pattern is anomaly and pattern

detection, in which machine learning is used in pattern finding with the data. This pattern is used

to find connections between information and determine if data is an outlier or can fit into a
ETHICAL IMPLICATIONS OF AI IN AUDITING 4

pattern. The application of pattern detection is the discovery of anomalies through intelligent

monitoring to make a difference in the detection of fraud. The sixth AI pattern is the

recognition that uses machine learning to identify desired information within data that is not

structured, such as video or audio. The seventh pattern of AI is goal-driven systems that use

machine learning to give people the ability to determine the solution to a given problem.

Fig 1: Artificial Intelligence patterns

The types of Artificial Intelligence

Artificial Intelligence has three types that pose various ethical implications in auditing

business; Autonomous AI, Augmented AI, and Assisted AI. Rice (2020) defines Assisted AI as

"a means of automating simple processes and tasks by harnessing the combined power of Big

Data, cloud, and data science to aid in decision-making." The function of Assisted AI is to help

people in the decision-making process (Munoko et al., 2020). Those who use Assisted AI are

required to have trust that the AI will perform the intended role. The main benefit of Assisted AI

is freeing basic tasks leaving the user with complex tasks.

The second type is Augmented AI, which allows people and organizations to perform

tasks they could not through the support of human decisions and not through independent

simulation of intelligence (Rice, 2020). Even though Augmented AI is more complex than

Assisted AI since it has the capability of making some decisions on its own, it is not, however,

independent of the user. Rice (2020) states that Augmented AI "suggests new solutions rather

than simply identifying patterns and applying predetermined solutions." People do not solely rely

upon the decisions made by Augmented AI since the user will be in charge of the evaluation of

the decisions Augmented AI makes. Munoko et al. (2020) found out that the user's autonomy is

the ethical risk posed by Augmented AI.

The final type of AI is Autonomous AI, the most advanced of the three types of AI. Rice

(2020) states that Autonomous AI is one "in which processes are automated to generate the

intelligence that allows machines, bots, and systems to act independently, independent of human

intervention." The sophistication nature of Autonomous AI gives it the ability to operate without

the interference of any user. An article by Munoko et al. (2020), Autonomous AI can "adapt to

their environments and perform tasks that would have been previously unsafe or impossible for a

human to do (e.g., the use of drones to perform inventory inspections autonomously of assets in

remote locations auditors do not have access to)." However, the ethical implication posed by

Autonomous AI is that people will not be in a position to find out how the system makes

decisions if it is left to operate independently.

The capability that AI intelligence has makes it different from any other software; it has

coded rules. An example of such software is Robotic Process Automation (RPA) which is

essential in performing various administrative tasks. RPA is "a technology that automates
ETHICAL IMPLICATIONS OF AI IN AUDITING 6

standardized and rule-based activities using scripts" (Gotthardt et al., 2020). RPA is a highly

process-driven system, unlike AI.

The Ethical Implications of AI in Auditing Business

Leidner and Plachouras (2017) define ethics as "the part of practical philosophy

concerned with all things normative (moral philosophy, answering the fundamental question of

how to live one's life) permeates all aspects of human action." Moral principles affecting people

and how they arrive at decision-making is the concept of ethics. Since AI is reliant on algorithms

designed by people and is programmed to make decisions, there are ethical implications with the

use of AI. When incorporating machine learning into the auditing business and making

decisions, ethical issues need to be considered. One example of ethical implication is the lack of

monitoring of machine learning since they use models that are predictive in auditing to help in

making predictions of the kind of transactions that may be fraudulent. Munoko et al. (2020)

identify that the lack of requirement by the new autonomous AI creates an ethical implication

because human interaction is no longer needed. Instead, available data is used to make

interpretations and overall predictions.

The outcomes from data populations that were previously used, also known as training

data, are used by machine learning tools and then implemented in learned training data rules to

the new data populations. The algorithms of machine learning allow an auditor to identify

anomalies in populations without the involvement of humans. An ethical implication in this type

of automation is the making up of decisions by AI that may not be linked directly to people due

to the complex nature of the decision. IESBA (2018) records that the code of ethics requirement

is "in complying with the principle of professional competence and due care, a professional

accountant shall take reasonable steps to ensure that those working in a professional capacity
ETHICAL IMPLICATIONS OF AI IN AUDITING 7

under the accountant's authority. Have appropriate training and supervision." The function of AI

is to find different elements of data relationships. While there may be prediction accuracy,

machine learning can also produce false positives. For instance, if there is a false positive

conclusion by an algorithm, and an individual tasked with reviewing the algorithm's results does

not seem to comprehend the process leading to the decision by AI, there will not be an

appropriate step taken by the user in fixing the problem. With this complexity, an auditor may

not be able to interact correctly with the system if there isn't proper training.

There are emerging technologies that may impact the decision-making process in the

auditing business; Natural Language Processing and facial Processing. These emerging

technologies have future importance to auditing business and ethical implications.

Facial Processing Technologies (FPT)

Raji et al. (2020) explain that FPT encompasses face detection, and the use of analysis

and recognition of face within FPT are for specific significant points of the face. There are a

variety of sectors where the application of FPT can be made, such as shopping venues, airports,

laptop and smartphone applications, and the law enforcement sector. Concerning research made

by Ouanan, Ouanan, and Aksasse (2016), a visual, non-verbal, and emotional channel is formed

by a facial expression that has a role in spoken communication support. FPT may be used in

auditing business during fraud audits and financial statements as a segment of the interview

process of fraud. Dickey et al. (2020) state that FPT can be of help in the interviews in

identifying facial patterns that may create a suggestion of deceit and nervousness. Auditing

business requirement is that there should be the need to inquire with the management through

fraud interviews. Using FPT can help auditors identify when the response from management

requires further investigations. Since there is training from the firms on fraud interviews,
ETHICAL IMPLICATIONS OF AI IN AUDITING 8

auditors are equipped with an understanding of behavioral patterns and expressions on the face

indicating the process of fraud.

Several ethical implications are related to FPT privacy, including the accuracy of FPT

and how the data used in training FPT is obtained. The main ethical implication is the absence of

federal standards and regulations surrounding FPT (Martin, 2019). With the absence of

established regulations, there may be a misuse of FPT. There are ethical standards that bind

auditors when using FPT. The images used in training the machine have their source questioned.

The challenge faced by auditing businesses is the development of suitable training data sets since

there is not enough training data access compared to companies like Facebook and Google (Issa,

Sun, & Vasarhelyi, 2016).

Natural Language Processing (NLP)

NLP is another aspect of AI that impacts auditing quality in the auditing business. NLP is

used to assist computers in comprehending the natural language of people (Garbade, 2018). The

development of NLP was to ease the user's work and have a natural communication with

computers. Similar to FPT, NLP has ethical implications in the auditing business. There are

possibilities of including pre-existing biases, presumptions, and prejudices. There are also

accuracy and privacy ethical implications for the NLP technology. There is always an inherent

risk of accuracy while using any algorithm and software that makes an interaction with humans.

There are also challenges in training NLP to interpret language as humans. Lack of transparency

is another ethical implication of NLP, which may come at the cost of privacy. While explaining

AI makes the way decisions, there might be a breach of privacy by the auditors. Munoko et al.

(2020) state that the created data laws protect people's privacy but not the data subjects, which

may experience the risk of inferential analytics.

Privacy and Data Protection

Buttarelli (2018) accounts that though data protection and privacy are not identical, about

ethics of AI use in auditing, the primary concern of privacy is privacy for information and data

protection, which can be understood as a way of protecting the privacy of information. There are

several ethical risks posed by AI based on machine learning. There is a need for large datasets,

on the other hand, for training purposes, and accessing such data can pose questions on the

protection of data. The ability of AI can be a problem since it may pose the risk of privacy, even

in a place where there is a lack of possibility of personal data access. There is also the

potentiality of AI in allowing anonymized personal data re-identification in unforeseen ways

before the apparent capabilities of machine learning. While there is the establishment of data

protection laws in several jurisdictions, AI's potentiality to create new risks of data protection

that have not been envisaged by legislation, thus creating new ethical implications for auditing

businesses. AI may worsen the situation through the generation or use of less widely employed

personal data like personal emotional data. Jagielski et al. (2018) indicate that though concerns

about data protection are a perennial problem in the entire ICT sector, AI may target new types

of security vulnerabilities like the attacks of model poisoning.

Therefore, issues of protection issues point out that the broader AI systems questions

reliability. While there is a more significant concern for all technical artifacts on reliability, the

machine learning system opacity and their lack of predictability mean that tradition's

deterministic testing regimes may not be applied to machine learning. The machine learning

system's output depends on the data quality used for training which may be hard to ascertain.

Security breaches can threaten the integrity of data in auditing and organizational or technical

aspects. Therefore, concerning this, machine learning systems' reliability may need an
ETHICAL IMPLICATIONS OF AI IN AUDITING 10

assessment from other system types, which may pose ethical implications for auditing businesses

if the system's outputs affect the value of ethics (Topol, 2019). By definition, machine learning is

not transparent about what other ICT systems may be. Transparency may be limited by the

commercial confidentiality of models and algorithms, where proprietary systems may be. In the

auditing business, lack of transparency may raise accountability questions and make it harder to

identify and address discrimination and biases questions.

Bias and Discrimination

CDEI (2019) cites that bias is another ethical implication of AI in the auditing business.

The challenge posed by machine learning systems is that they can inadvertently or intentionally

bring up the already existing reproduction of biases. Alongside the biases brought up in the

auditing business, other examples may include perpetuated biases in the probation process

through machine learning or machine learning biases on gender recruitment (Raso et al. 2018).

Discrimination stands out as an ethical implication of AI in the auditing business. Still, it has

been for a long time recognized as an infringement of human rights, which is illegal in several

jurisdictions. As there is a risk of infringing human rights by AI, the highlight has been focused

on the infringement of the right to quality by machine learning in the auditing business (Access

Now Policy Team, 2018).

In systems that interact directly with human beings in the physical world like critical

healthcare institutions and financial institutions or autonomous vehicles like Tesla (BmVI,

2017), safety is one of the critical ethical implications.

Recommendations to Manage Risks Related to Ethical Use of AI

Explainable Algorithms Need – Algorithms for machine learning are only as better than

the provided training data. These system users can take the quality of data for granted, which

may result in over-trusting the algorithms' predictions. In addition, some machine learning

models are hard to interpret, making it difficult to understand the nature at which decisions were

made (in other terms known as black box decisions). Another ethical concern is when the data

with low quality embeds bias is implemented in models that are difficult to interpret, thus

making interpreting bias harder. Researching into the interpretability of machine learning models

will help build trust in artificial intelligence. More broadly, providing education to stakeholders

regarding AI would make it easy to increase significant benefits and digital literacy. The

collaboration of education industries and government institutions can help explain how sound

models and data can both be significant to the ethical use of AI.

Application of ethics within a cost-benefit framework – AI ethics should be applied in

specific contexts and levels of understanding. The application of ethics is through a cost-benefit

framework would allow comparison of risks associated with AI by agencies (for instance, loss of

funds, discrimination, and potential for human danger) together with the benefits like saving

funds, egalitarian treatment, and saving lives, throughout the operations and development of

algorithm's lifecycle. Confidence intervals and scenario planning is always included in cost-

benefit analyses, which may be significant in AI systems evaluation over time – only if the 'costs'

which are considered not only include financial costs that are quantifiable but intangible and

value-based risks too (e.g., avoiding privacy or bias harms). If correctly implemented, this

approach may provide clear communication of risks and decisions on when and how AI can be

used. 
ETHICAL IMPLICATIONS OF AI IN AUDITING 12
ETHICAL IMPLICATIONS OF AI IN AUDITING 13

Conclusion

AI has numerous uses in the auditing business. Technologies like NLP and FPT are some

of the promising technologies for the auditing business. Some problems are not possible to solve

while using traditional approaches, but AI can be incorporated to solve these problems. Using

these technologies in the decision-making process of auditing, such as fraud interviews, are

essential in improving the general quality of audit in auditing businesses. Implementing AI

technologies has a significant risk for the auditing profession if not considered. Ethical

implications such as biases, transparency, data protection, confidentiality, accuracy, and privacy

are some of the AI problems that need to be dealt with when developing the technology. For

auditors to be in a position to gain and maintain the public's trust, they should be able to

understand the ethical issues surrounding the AI technology before its implementation.
ETHICAL IMPLICATIONS OF AI IN AUDITING 14

References

Access Now Policy Team (2018) The Toronto declaration: protecting the right to equality and

non-discrimination in machine learning systems. Access Now,

Toronto. https://www.accessnow.org/cms/assets/uploads/2018/08/The-Toronto-

Declaration_ENG_08-2018.pdf.

BmVI (2017) Ethik-Kommission: Automatisiertes und vernetztes Fahren. Bundesministerium für

Verkehr und digitale Infrastruktur,

Berlin. https://www.bmvi.de/SharedDocs/DE/Publikationen/DG/bericht-der-ethik-

kommission.pdf?__blob=publicationFile

Buttarelli G (2018) Choose humanity: putting dignity back into digital. In: Speech at 40th

international conference of data protection and privacy commissioners,

Brussels. https://www.privacyconference2018.org/system/files/2018-10/Choose

%20Humanity%20speech_0.pdf.

CDEI (2019) Interim report: Review into bias in algorithmic decision-making. Centre for Data

Ethics and Innovation, London. https://www.gov.uk/government/publications/interim-

reports-from-the-centre-for-data-ethics-and-innovation/interim-report-review-into-bias-

in-algorithmic-decision-making

Davenport, T. H., Raphael, J. (2017). Creating a Cognitive Audit. CFO. Retrieved from:

https://www.cfo.com/auditing/2017/07/creating-cognitive-audit

Garbade, M. J. (2018). A Simple Introduction to Language Processing. Medium. Retrieved from:

https://becominghuman.ai/a-simple-introduction-to-natural-language-

processingea66a1747b32
ETHICAL IMPLICATIONS OF AI IN AUDITING 15

Gotthardt, M., Koivulaakso, D., Paksoy, O., Cornelius, S., Martikainen, M., Lehner, O. (2020).

Current State and Challenges in the Implementation of Smart Robotic Process

Automation in Accounting and Auditing. ACRN Journal of Finance and Risk

Perspectives, 9, 90-102. https://doi.org/10.35944/jofrp.2020.9.1.007

IESBA. (2018). Handbook of the International Code of Ethics for Professional Accountants.

New York: International Federation of Accountants, Professional Code. 110, 20

Issa, H., Sun, T., 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), 1–20. https://doi.org/10.2308/jeta-10511

Jagielski M, Oprea A, Biggio B et al. (2018) Manipulating machine learning: poisoning attacks

and countermeasures for regression learning. In: 2018 IEEE symposium on security and

privacy (SP), San Francisco. https://doi.org/10.1109/SP.2018.00057

Kokina, J., Davenport, T.H. (2017). The emergence of artificial intelligence: How automation is

changing auditing. Journal of Emerging Technologies in Accounting, 14(1), 115-122.

https://doi.org/10.2308/jeta-51730

Leidner, J. L., Plachouras, V. (2017). Ethical by Design: Ethics Best Practices for Natural

Language Processing. Association for Computational Linguistics, 30-40.

https://doi.org/10.18653/v1/W17-1604

Martin, N. (2019). The Major Concerns Around Facial Recognition Technology. Forbes.

Retrieved from: https://www.forbes.com/sites/nicolemartin1/2019/09/25/the-major-

concerns-aroundfacial-recognition-technology/?sh=74e682224fe3
ETHICAL IMPLICATIONS OF AI IN AUDITING 16

Munoko, I., Brown-Liburd, H. L., Vasarhelyi, M. (2020). The Ethical Implications of Using

Artificial Intelligence in Auditing. Journal of Business Ethics, 167, 209-234.

https://doi.org/10.1007/s10551-019-04407-1

Ouanan, H., Ouanan, M., Aksasse, B. (2016). Facial Landmark Localization: past, present, and

future. 2016 4th IEEE International Colloquium on Information Science and Technology

(CIST), 487-493. https://doi.org/10.1109/CIST.2016.7805097

Raji, I. D., Gebru, T., Mitchell, M., Buolamwini, J., Lee, J., Denton, E. (2020). Saving Face:

Investigating the Ethical Concerns of Facial Recognition Auditing. Proceedings of the

AAAI/ACM Conference on AI, Ethics, and Society, 145-151.

https://doi.org/10.1145/3375627.3375820

Raso FA, Hilligoss H, Krishnamurthy V et al. (2018) Artificial intelligence & human rights:

opportunities & risks. Berkman Klein Center Research Publication No. 2018-6.

https://doi.org/10.2139/ssrn.3259344

Rice, D. (2020). Assisted Intelligence vs. Augmented Intelligence and Autonomous Intelligence.

Fed Tech. Retrieved from: https://fedtechmagazine.com/article/2020/01/assisted-

intelligence-vs-augmentedintelligence-and-autonomous-intelligence-perfcon

The Seven Patterns of AI. (2019). Cognilytica. Retrieved from:

https://www.cognilytica.com/2019/04/04/the-seven-patterns-of-ai/

Topol EJ. (2019) High-performance medicine: the convergence of human and artificial

intelligence. Nat Med.  2019;25:44–56. https://doi.org/10.1038/s41591-018-0300-7.

Walch, K. (2019). The Seven Patterns of AI. Forbes. Retrieved from:

https://www.forbes.com/sites/cognitiveworld/2019/09/17/the-seven-patterns-ofai/?

sh=b7f05af12d01