You are on page 1of 24

GROUP 11

The impact of natural language processes ChatGPT in


Audit

Kudakwashe Maunga R204062L


Felicia R Mupanduki R204072A
Fatima C Mwanyali R204056G
Tinotenda Marakurwa R204067R
Grace Mazorodze R203654T
INTRODUCTION

The evolution of technology is taking place at a fast pace, and knowingly or unknowingly everyone is
generating data in some or the other way.
The purpose of this article is to explore on natural language processes and the use of ChatGPT in
auditing firms.
1

Researchers have defined natural language processing as a subfield of artificial intelligence which focuses
computational linguistics interpretation. Natural language can be counted in different forms which include
speech, singing, facial expressions and body language. This field encompasses several areas of textual and
audio interpretation by integration of machine learning methods which behaves statistically.
ChatGPT in Auditing

• Another Artificial Intelligence (AI) software this study is going to look at is called Chat Generative Pre-
Trained Transformer (ChatGPT), which has gained recognition for its ability to improve conversational
and written tasks (Deng and Lin, 2022) and its potential to revolutionise the business world, including
external auditing.

• The Big Four audit firms are currently investing billions of dollars in AI and data analytics technology
(Kapoor, 2020; Maurer, 2021), intending to improve audit efficiency, effectiveness, and decision-making
(Bloomberg Tax, 2020).
OBJECTIVES AND RESEARCH QUESTIONS
• Objectives

• This paper examines Natural language processes and ChatGPT


• Research Questions
literature in auditing. Our objectives are:
 How does natural language processing contribute to the
 To analyse the contribution of natural language processes in auditing process in financial institutions?
financial institutions and its impact in auditing large volumes of
 What are the key challenges in implementing natural language
data processing techniques for auditing large textual data and how it

 To identify the key challenges in implementing natural language can be used to detect data?

processing techniques for auditing large volumes of textual data  How can ChatGPT be used to streamline the auditing process
in financial institutions and the potential privacy in using
 To identify and synthesize the literature on the use of ChatGPT
ChatGPT in auditing processes?
in financial institutions and the potential privacy concerns for
 How can ChatGPT be integrated with the existing auditing
auditing processes.
software’s to automate data analysis and the key challenges in
 To explore on the ways ChatGPT can be integrated with implementing ChatGPT for auditing purposes?
existing audit software’s to automate data analysis and detect
fraud

 To identify the key challenges in implementing ChatGPT for


auditing purposes in institutions and mitigation measures
The remainder of this paper is organized as follows. Section 2 describes the background and evolution of NLP
and ChatGPT. A detailed synthesis of the NLP and ChatGPT literature literature will be discussed in section 3.
Methodology used to identify the relevant follows in chapter 4. Our section 5 will present the results and
findings from the researches that will be conducted to answer the research questions and give an overview of the
entire article as we conclude in the last part. identifying future research opportunities as well as likely obstacles
and provide concluding observations.
Background
 Generally, the history of NLP is thought to have started in the 1950s 1. NLP has gone through a fast developing period during the last decade and is
widely adopted in various industries today, including audit. Four major applications of NLP that auditors tend to consider mainly, text
classification, information retrieval, Natural language generation and Natural language understanding
 Before NPL, vouching, or examining documentary evidence to verify the accuracy and occurrence of a transaction, a basic audit practice and
probably also one of the most tedious ones was unsystematic. Now, with the capability of optical character recognition (‘OCR’) to convert hard
copies into machine-readable formats and NLP to retrieve key information from documents, such as invoices and delivery orders, it is possible to
automate the vouching task and free up auditors’ time for higher value tasks which eliminate human errors and increase the accuracy of data entry.
 Studies has revealed that NLP’s text introduced classification ability and today, highly accurate text classification model (Currently, text
classification models can achieve an accuracy higher than 90%) is therefore playing a key role in automating the process of filtering for negative
news. NLP came with benefits of process textual information in any language, which saves translation time for the auditor or reduces the cost of
employing a professional translator.
 Earlier NPL processing was not trained to identify meaning of phrases or sentences in text which made the analysis of data captured time
consuming and exhausting. Throughout the years continued researches provided and contributed to NLU understanding the line of a business for a
client to be in a position to best advise them on key contributions towards growth in their competitive industries.
Natural Language Processing
 NLP proved to have potentially transformed the audit industry, allowing auditor’s opinion to be based on accurate facts as less effort is now used
on analysing complex texts explaining client’s line of business and industry. Moreover NLP tools such as NLU came with the provision of the
ability to process more than 80% of unstructured data formats with processing parity reducing repetitive tasks hence improving audit efficiency
in vouching and performing recalculations.
 NLP has played a major role in automating audit procedure determining the accuracy, completeness, occurrence and test of control processes
improving efficiency and allowing auditors provide opinions that are satisfactory in short deadlines.
 Moreover, the introduction of ChatGPT and the fundamentals of IT auditing, responses were now generated by predicting, on statistical patterns,
the most likely continuation of the conversation based on the patterns it has learned during training through understanding conversation based
human language texts and provide analysed data texts and are more accurate and short yet meaningful. It has improve auditors’ processes and
control mechanisms regarding various tasks performed in determining the compliance of a clients’s principal activities and operations to
governing laws in a more systematic way considered the type of organisation involved is mostly computerised, which improving efficiency in
meeting deadlines.
ChatGPT was delevoped as a mechanism to manage risk in IT audit. To mitigate these risks, it is important
to integrate ChatGPT in IT auditing with a cautious and critical mindset. The ChatGPT has been offering
significant opportunities for IT auditors, including enhanced efficiency, real-time monitoring, improved
data analytics and risk assessment..

• ChatGPT was delevoped as a mechanism to manage risk in IT audit. To mitigate these risks, it is
important to integrate ChatGPT in IT auditing with a cautious and critical mindset. The ChatGPT has
been offering significant opportunities for IT auditors, including enhanced efficiency, real-time
monitoring, improved data analytics and risk assessment..
LITERATURE REVIEW
Literature review is a type of academic writing that provides an overview of existing knowledge on a
particular topic. The purpose of a literature review is to identify key concepts, theories, and findings from
existing research and to explore the relationships between them. Over the years, NLP has witnessed
tremendous advancements; however, several ongoing research areas and identified gaps hinder its full
potential. This literature review aims to explore these gaps, identify their implications, and propose potential
mitigation strategies.
In the article "Advances in natural language processing,(" Julia Hirschberg and Christopher D Manning )
discuss the recent advancements in the field of natural language processing (NLP). NLP is a subfield of
artificial intelligence that focuses on the interaction between computers and human language. The authors
begin by highlighting the increasing importance of NLP in various applications such as search engines,
machine . One significant advancement in NLP is the use of statistical techniques, especially machine
learning, to improve language models. These models enable computers to understand and generate human
language more accurately. The authors also discuss the development of deep learning techniques, which
have further enhanced the performance of NLP systems by allowing them to learn hierarchical
representations of language.
There are also several scholars who have discussed the application of natural language processing (NLP) in the
field of auditing. Here are a few
1. Kogan et al. (2019) examined how NLP techniques can enhance auditors' ability to identify misstatements in
financial statements. They proposed the use of NLP algorithms to analyze and compare textual information
from various financial documents to detect inconsistencies and potential reporting irregularities. Kogan
highlighted the benefits of applying NLP in extracting relevant audit evidence from multinational corporations
financial reports. They suggest that NLP can facilitate the identification of potential of red flags and
irregularities.
2. Krahel and Titera (2020) explored the use of NLP in analyzing audit reports and extracting valuable
information from unstructured data. They emphasized the potential of NLP to automate the extraction of audit
findings, recommendations, and management responses, thereby facilitating the analysis and benchmarking of
audit quality.
There is limited research specifically focused on the application of natural language processing (NLP) in
relation to international, regional, and geographical aspects. However, scholars have discussed the broader
applications of NLP in various domains that may indirectly relate to these aspects. Here are a few perspectives
1. Piao et al. (2019) examined the use of NLP for sentiment analysis in the context of international relations.
They explored how NLP techniques can be used to analyze large volumes of textual data, such as news
articles and social media posts, to understand public sentiment towards specific countries or regions.
2. Zhang et al. (2018) discussed the potential of NLP in cross-lingual information retrieval, which allows users
to search for information across different languages. They highlighted the importance of NLP techniques in
overcoming language barriers and facilitating international knowledge sharing.

However there are gaps found in these articles which needs to be mitigated .
There are several gaps and missing information in the field of natural language processing (NLP) that many
scholars haven't extensively covered. Some of these include:
-Multilingual and cross-lingual understanding: Most NLP research has focused on English language
processing, leaving a significant gap in the development and application of NLP techniques for other
languages. Achieving effective multilingual and cross-lingual understanding remains a major challenge, as
languages vary greatly in structure, grammar, and vocabulary.
- Common sense reasoning: Human language often relies on common-sense knowledge and reasoning,
which is not explicitly expressed in text. NLP systems struggle to infer and apply this background knowledge,
limiting their ability to understand language in a way that humans do effortlessly.
-Understanding longer and complex texts: Current NLP techniques excel at processing short texts, but
struggle with longer and more complex documents. Effective comprehension of lengthy documents, such as
scientific papers or legal texts, requires advancements in summarization, information extraction, and
document understanding.
-Privacy and security: With the increasing use of NLP in various applications, concerns arise regarding user
privacy and data security. Research is needed to address these issues and develop robust methods for
preserving privacy and securing sensitive information in NLP systems.
Addressing these gaps will require interdisciplinary collaboration and continued research efforts to push the
boundaries of natural language processing and develop more sophisticated and capable NLP systems.
MITIGATING STRATEGIES

- Large-scale Datasets and Pretraining: Increasing the availability of high-quality datasets and using pretrained
models can help improve NLP performance and address issues related to lack of data.
- Explainable AI: Developing NLP models that can explain their decisions and provide interpretable output can aid
in building trust and ensuring transparency in NLP applications.
- Cross-cultural and Multilingual Training: Incorporating diverse cultural perspectives and multilingual training can
enhance the generalization capabilities of NLP models across different languages and cultures.
METHODOLOGY

INTRODUCTION
NLP is a branch of artificial intelligence that deals with the interaction between computers and human
language. In the field of auditing, where large amounts of financial data need to be analyzed and interpreted,
NLP techniques offer potential benefits in terms of efficiency, accuracy, and risk identification.
The research aims to investigate how NLP can be applied in auditing to enhance the processing and analysis
of large volumes of textual data, such as financial reports, invoices, contracts, emails, or other financial
documents
Literature section and Research approach

Literature Section
• We began our Research by performing key word searches of databases, including, the primary search terms
included natural language processing (NPL), text mining, text analysis, Artificial Intelligence (AI),
Computerized Content Analysis. We accessed about 4 journals and 1 dissertation
Research Approach
• There are three main ways in which research approaches can be modified to suit a particular study and these
are qualitative, quantitative and mixed approach, For the purpose of this study, the researcher resorted to the
use of a qualitative research approach. qualitative research can offer rich, comprehensive data that can be
examined to find themes and patterns.
Philosophical assumption and Research design
Philosopical assumption
• The common research philosophies are positivism, realism, interpretivism, subjectivism, and pragmatism. The current
study chooses interpretivism due to the literature review’s discussion of the contributions of Natural language
processing in audit process of the financial institutions and the impacts in auditing large volumes of data. The purpose
of research we used interpretivism.

Research Design
• A descriptive research design has been selected for this study. The main advantage of a descriptive research design is
that it provides an accurate picture of the Natural Language processing contributions to the auditing process in financial
institutions, also to analyse the impacts of Natural language processing in auditing large volumes of data and the impact
of ChatGPT on auditing processes.
Data collection technique
• Secondary data have been considered in data collection. . The main source used in the study was the scanning
bibliographies of 4 literature reviews that surveyed NLP developments in accounting, audit and finance domains. We
accessed about 4 journals and 1 dissertation. The main advantage of secondary data is that it is readily available and
cost efficient
Research instrument, Ethical considerations,
Population and Sample

Research instrument and


Ethical consideration Population and Sample
Research instrument Population

• The researcher used literature review tool or • The population for this study includes professionals involved
bibliographic tool. The researcher used these to gain in auditing large volumes of financial data, such as auditors,
information from different sources such as journals and accounting professionals, data analysts, and researchers in
also dissertation. The researcher used this tool because its the field of auditing
time efficiency since we were obtaining secondary data Sample
Ethical consideration • In this study we used two samples, which are Researchers
and academics with expertise in auditing, accounting, or
• were also put in place through the research process to
natural language processing which helped to gather insights
ensure privacy and confidentiality of the data collected.
from a scholarly perspective
We also considered the potential bias in the data when
analyzing it.
FINDINGS ON NLP AND CHATGPT
• NLP and ChatGPT promotes automation of routine tasks and processes such as data extraction, data
validation and report generation that auditors manually performed previously. It allows auditors to focus on
more complex and critical aspects such as data evaluation thereby freeing the auditors time from repetition
of tasks. Automation of the audit process improves overall efficiency, eliminates human errors and auditors
will be able to test against full populations which presents more accurate results due to all of the data
being analysed as compared with the traditional approach of testing a sample of the data.
 NLP and ChatGPT can assist auditors in risk assessment as they analyse large volumes of data such as
systems logs, financial records, network traffic, and identification of patterns and anomalies associated
with organization systems and processes. Thus providing the auditors with a deeper understanding of an
organisation’s financial records and potential risks or irregularities associated with an organisation. These
valuable insights allow auditors to focus their testing efforts on high –risk areas where material
misstatements are more likely to occur thereby leading to a more targeted and effective audit.
 NLP and ChatGPT improves accuracy and consistency. ChatGPT can be used as a real-time monitoring tool
that will continuously analyse data streams to generate alerts for potential security breaches or unauthorized
access attempts. This proactive approach enables auditors to respond promptly to emerging threats and mitigate
risks before they escalate. Natural Language Processing helps auditors in the documentation of procedures,
findings and recommendations. It is able to generate concise and accurate reports based on the analysis of both
structured and unstructured sources of audit data leading to consistency and efficiency in the reporting process.
 ChatGPT can serve as a knowledge respiratory, storing best practices, industry standards and regulatory
guidelines. It can provide auditors with instant access to relevant information and assist in training new
auditors by answering queries and providing guidance based on past experiences
 Natural Language Processing algorithms can analyse transactional data to identify potential fraudulent
activities or irregularities. By flagging suspicious patterns, NLP helps auditors focus their attention on high-risk
areas, leading to more effective fraud detection
DRAWBACKS OF NLP AND CHATGPT IN AUDIT PROCESSING
• Over-reliance on NLP and ChatGPT can inhibit critical thinking. Auditors need to maintain a balanced
approach and not completely rely on ChatGPT and NLP for critical decision making. They should exercise
their professional judgement. Validate outputs and supplement them with their own expertise to ensure accurate
and appropriate conclusions.
• Lack of contextual understanding. Auditors should exercise caution and validate ChatGPT’s output with human
judgement.
• Inaccurate or incomplete information. Responses from ChatGPT may be generated without fully understanding
the specific context of the audit engagement.

• ChatGPT does not have the facility of embedding with legal and compliance regulatory laws. This is risk in
auditing since auditor would want to protect sensitive and confidential information. There is exposure of
sensitive information to third parties leading to data misuse.
• Limited domain expertise. NLP may lack specialised knowledge and industry-specific expertise that auditors
possess which can impact complex audit scenarios
CONCLUSION

In conclusion, NLP and ChatGPT offer significant opportunities in the auditing process such as improved
accuracy and consistence, enhanced risk assessment, enhanced fraud detection, and workflow automation.
However, auditors must understand ethical concerns, security risks, biases and limitations associated with NLP
and ChatGPT. ChatGPT and Natural Language Processing should aid the audit process and not replace the
expertise and judgment of human auditors. Moreover, it is essential to work closely with cyber security
professionals and follow established security frameworks and standards, such as ISO 27001, COBIT 2019 and
NIST Cyber security Framework, to ensure a comprehensive and robust security posture. Regularly assessing
and improving security controls to maintain the confidentiality and integrity of sensitive data should be
considered to establish trust and confidence with AI technology. According to Salijeni et al. (2018), there is
limited empirical evidence on the effects of emerging technology on the audit process and profession. Future
research could add value by performing a quantitative analysis to determine the validity of an individual threat
to audit firms and the auditing profession. As no attempt has been made to determine whether the threats
outweigh the benefits created or vice-versa, and no conclusion has been formed as to whether audit firms
should engage in technological implementation, this presents scope for future research.
REFERENCES

Deloitte, (2020). Advancing Audit Quality with Smarter Audits [Online]. Available at: https://
www2.deloitte.com/za/en/pages/audit/solutions/smarter-audits.html.

Gomez, J.C.M., Schumann G. (2021). Natural Language Processing in Internal Auditing- A Structured
Literature Review.

Hasan, A. (2022) Artificial Intelligence (AI) in Accounting & Auditing: A Literature Review. Open Journal
of Business and Management, 10, 440-465. 10.4236/ojbm.2022.101026.

Ingrid E. Fisher, Margaret R. Garnsey, Mark E. Hughes (2016, March 01). Natural Language Processing in
Accounting, Auditing and Finance: A Synthesis of the Literature with a Roadmap for Future
Research.https://onlinelibrary.wiley.com/doi/10.1002/isaf.1386

KPMG, (2020). Audit Innovation [Online]. Available at:


https://home.kpmg/us/en/home/services/audit/auditinnovation.html.
Martinez, R. (2019). Artificial Intelligence: Distinguishing between Types & Definitions. Nevada Law Journal,
19, Article 9.
https://scholars.law.unlv.edu/nlj/vol19/iss3/9

PwC, (2019). Harnessing the power of AI to transform the detection of fraud and error [Online]. Available at:
https://www.pwc.com/gx/en/about/stories-from-across-the-world/harnessing-the-power-of-ai-totransform-the-
detection-of-fraud-and-error.html.

Salijeni, G., Samsonova-Taddei, A. and Turley, S. (2018). Big Data and changes in audit technology:
contemplating a research agenda. Accounting and business research, 49(1), pp. 95–119.

Zhang, C. (2019). Intelligent Process Automation in Audit. Journal of emerging technologies in accounting,
16(2), pp. 69– 88

You might also like