The Use of Artificial Intelligence in Banking Industry

Reza Farishy

Master of Management, Airlangga University, Indonesia

E-mail: [email protected]

Keywords

ABSTRACT

artificial intelligence, banking industry, credit rating, bank failure, threat

Industry 4.0, also known as the fourth industrial revolution, has altered society and the economy by introducing intelligent robotics, artificial intelligence (AI), cloud computing enormous data sets, the Internet of Things (IoT), and 3D printers, among other scientific advances. To maintain competitiveness and keep up with global competition, it is vital to adapt to modern technology. The financial sector is a vibrant market with intense competition for products and services, and advancements in information technology have led to the development of highly valuable new technologies. This essay addresses the possible advantages of artificial intelligence in the banking sector. The study utilized a Systematic Literature Review (SLR) to evaluate the current literature on AI in the banking industry. The results of the SLR demonstrate that AI has been utilized in the banking industry in a variety of ways, including credit rating models and bank collapse prediction. In establishing credit card eligibility, logistic regression models were shown to be effective, with an accuracy rate of 80.43 percent. With a precision rate of 75.7% and a recall rate of 75.7%, artificial neural networks (ANNs) were shown to be the most accurate method for predicting bank collapse based on financial characteristics.  Overall, the study indicates that AI has the ability to dramatically improve the banking business by enhancing efficiency, precision, and decision-making procedures. The study has limitations and potential biases, including the exclusion of non-English language articles and the possibility of a selection bias. To explore the full potential of AI in the banking business, additional study is required.

INTRODUCTION

The fourth industrial revolution, also called Industry 4.0, is a transformation of the economy and society in combination with intelligent robotics, AI, cloud computing, the Internet of Things (IoT), and other scientific breakthroughs (Hassoun et al., 2022). To able to stay competitive and avoid going out of business because of global competition and the problem's growing complexity, there is a need to adapt to current technology (Kotabe & Helsen, 2022). In Industry 4.0, every work is involving technology, including accounting, distribution, marketing, supply chain, etc (Fig 1). It needs a lot of time and effort to work on such big data without an integrated system that helps user to process the data. Automation can be produced by IoT devices by allowing them to connect with one another, gather data, and take actions depending on that data(Munirathinam, 2020). Thus, every work involving automation system and technologies is always more effective and efficient.

Figure 1 Business Employing AI (www.statista.com)

Intense rivalry exists for products and services in the financial sector [1]. The first ATM, SMS banking, m-banking, and internet banking all faced rivalry in the past(Dzombo, Kilika, & Maingi, 2017). With the introduction of the Internet in the late 1990s, the realization of financial transactions began(Machkour & Abriane, 2020). The development of smartphones, however, marks the true acceleration of access to the Internet(Xie, Zhang, & Zhu, 2017). The creation of new technologies that are highly regarded in the financial industry is the result of technology advancements(Bisht et al., 2022). The usage of QRIS payment, internet banking, and virtual account are a few examples on how information technology enhances transaction process in banking industry(Anderson et al., 2019).

The most important aspects of this essay are its discussion of how artificial intelligence might benefit banking industry.

METHODS

The present literature on artificial intelligence (AI) in the banking industry was examined in this study using a systematic literature review. The research process is ilustrated in Fig 2. The research conducted a detailed and methodical process to discover, select, and assess relevant research from a variety of sources, including academic journals, conference proceedings, and other scholarly publications. Many electronic databases, including Web of Science, Scopus, and Google Scholar, were utilized to conduct the search.

For the research to be considered, inclusion criteria were established, such as a focus on AI in the banking industry and publication in English-language journals or conferences. Exclusion criteria, such as studies that did not fit the inclusion requirements and those that were duplicates or irrelevant to the topic, were also developed. Following the initial search, a screening procedure was done to assess the relevancy of each study based on its title, abstract, and keywords. The studies that met the inclusion requirements were then subjected to a full-text analysis, and their quality was evaluated using predetermined criteria, such as the relevance of the research issue, the validity of the methodology employed, and the credibility of the results.

Data extraction was undertaken to collect and evaluate pertinent data from the selected publications, such as the publication year, the study design, the research question, the methodologies employed, and the major findings. The review results were aggregated and presented in narrative format to give examination of the current state of AI research in the banking business. The limitations of the study were also acknowledged, including the possibility for selection bias and the exclusion of non-English language publications.

Figure 2. Research method and data extraction

RESULT AND DISCUSSION

By connecting people, machines, and technology, Industry 4.0 has changed society's socioeconomic condition in addition to its impact on many industries [7]. In order to present the information gleaned from the study of the papers, this section discusses AI-based approaches that have been used by practitioners in banking industry from various business process and researchers.

AI on Credit Rating Models

Wu et al., 2021 conducted research using dataset which was sourced from china machine learning databases. It contained financial data on a sizable bank client that had been made anonymous. Of these, 517 had credit cards and were creditworthy, whereas 183 had credit card debt. One categorical feature, education, and seven numerical attributes, including age, seniority, residency duration, income, debt to income ratio, credit card debt, and other debt, were included in this data collection. The descriptive statistics of the data presented on the Table 1.

Table 1. Descriptive Statistics of Input Variables

Sources: Wu et al., 2021

Table 2. Result on Traditional AI Models

Sources: Wu et al., 2021

The study assessed the effectiveness of numerous classic AI models, such as logistic regression, SVM, and MP, in determining credit card eligibility. The model with the highest accuracy, as shown in Table 2, was logistic regression, with an accuracy of 80.43 percent. SVM and MP followed with accuracy values that were lower.

AI on Predicting Bank Failure

Le & Viviani (2017) collected 31 ratios from the bank financial statements using a sample of 3,000 institutions, consisting of 1,438 inactive and 1,562 active banks over a 5-year period data. Table 3 provides more information regarding the extracted ratios. The researchers then utilized multiple machine learning algorithms to predict bank failure, including ANNs, k-NN, LR, SVM, and LDA.

Table 5 indicates that ANNs beat all other methods, with a precision rate of 75.7% and a recall rate of 75.3%. k-NN and LR both yielded conclusions with a precision of roughly 74%, however SVM and LDA fared badly, attaining only 71.6% and 72% precision, respectively. Notably, the TP rate was 75.3% while the FP rate was 25.9%.

Overall, the study indicates that ANNs are the most accurate method for predicting bank failure based on financial parameters, and they may have applications in the banking business.

Table 3. Measured Financial Ratio

Source: Le & Viviani, 2017

Table 4. Descriptive Statistics for the 31 Financial Ratios

Source: Le & Viviani, 2017

Table 5 showed that, No matter how you measure performance (75.7% accuracy and 75.3% recall), ANNs outperformed all other approaches. As can be observed, k-NN and LR both produced findings with approximately 74% precision. SVM and LDA perform badly, with just 71.6% and 72% accuracy, respectively.

Table 5. Experiment Result

Source: Le & Viviani, 2017

On Cyber Threat Detection

Noor et al study.'s from 2019 looked at the effectiveness of several machine learning models for identifying cyberthreat actors (CTAs). The 36 threat actors that have been identified during 2012–2018 are described in 327 thread reports from 26 sources, according to the researcher (see Table 6). The experiment result shown in Table 7 which shows the total values that was calculated by averaging on each CTA types. Later researcher found that DLNN is the best tools to accurately predict threat detection (94% accuracy and 90% precision). k-NN was the most inaccurate tools for the threat detection.

The research found that the naive bayes, RF, and DLNN models could successfully detect CTAs with good accuracy, precision, recall, and f-measure. Feature selection enhanced the Naive Bayes model's precision, recall, and f-measure but did not appear to enhance the models' attribution accuracy. Nonetheless, the accuracy, recall, and f-measure decreased with RF and DLNN models.

The study reported low accuracy, precision, recall, f-measure, and false-positive rate when evaluating the models using the ATT&CK dataset (the Adversarial Tactics, Techniques, and Common Knowledge). This may be related to the restricted quantity of data accessible in the dataset, as there was only one occurrence of a high-level indicators dataset for each CTA. The study was therefore unable to analyze the attribution of cyber assaults using cross-validation methodologies. Overall, the study emphasizes the significance of analyzing machine learning models using many datasets to determine their efficacy in identifying and attributing cyber intrusions. It also indicates that feature selection may not necessarily increase the accuracy of machine learning models' attribution.

Table 6. Summary of CTA

Source: Noor et al., 2019

Table 7. Experiment Result

Sources: Noor et al., 2019

CONCLUSION

Industry 4.0 has had a substantial impact on the socioeconomic state of society and has revolutionized a variety of industries [10]. The financial industry is not an exception; practitioners and researchers have utilized AI-based technologies to enhance numerous business operations. The artificial intelligence movement has greatly improved e-finance, as previously manual or statistical model-based tasks have evolved into more intelligent, autonomous, and predictive ones

The research covered in this paper shows the efficacy of several machine learning algorithms in the banking sector based on previous research. Wu et al. (2021) reported that logistic regression was the most efficient approach for assessing credit card eligibility, whereas Le & Viviani (2017) found that artificial neural networks (ANNs) were the best accurate method for forecasting bank failure based on financial characteristics. Noor et al. (2019) discovered that naïve bayes, RF, and DLNN models is a moderated tools to detect cyber threat actors.

It is vital to remember, however, that accuracy is only one of several metrics used to measure the effectiveness of a machine learning model, and it may not always provide a comprehensive view of the model's usefulness. Depending on the application, other metrics such as precision, recall, and F1 score may also be significant. Moreover, the generalizability of the conclusions may be constrained by the specific dataset used in the study.  Overall, these studies illustrate the potential for AI-based approaches to improve decision-making and increase efficiency in the banking business. AI will undoubtedly play a larger role in the banking industry as technology continues to evolve, allowing financial organizations to better manage risk, improve customer experience, and increase profitability.

This study has identified a number of frameworks that have been put out by business professionals for AI integration. As a result, a uniform framework is needed that industrial practitioners from various fields may utilize to incorporate AI into their industrial processes. Despite the study's concentration on AI/ML, there is still a lot of ground to cover in the vast topic of Industry 4.0, which cuts across many other industries. The primary objective of this study, which was exploratory in nature, was to give a comprehensive review of the technologies and research fields related to banking-related artificial intelligence in the context of Industry 4.0 to non-specialists.

REFERENCE

Anderson, Ross, Barton, Chris, Bölme, Rainer, Clayton, Richard, Ganán, Carlos, Grasso, Tom, Levi, Michael, Moore, Tyler, & Vasek, Marie. (2019). Measuring the changing cost of cybercrime.

Bisht, Deepa, Singh, Rajesh, Gehlot, Anita, Akram, Shaik Vaseem, Singh, Aman, Montero, Elisabeth Caro, Priyadarshi, Neeraj, & Twala, Bhekisipho. (2022). Imperative role of integrating digitalization in the firms finance: A technological perspective. Electronics, 11(19), 3252.

Dzombo, Gift Kimonge, Kilika, James M., & Maingi, James. (2017). The effect of branchless banking strategy on the financial performance of commercial banks in Kenya. International Journal of Financial Research, 8(4), 167–183.

George, A. Shaji, & George, A. S. Hovan. (2023). A review of ChatGPT AI’s impact on several business sectors. Partners Universal International Innovation Journal, 1(1), 9–23.

Goodell, John W., Kumar, Satish, Lim, Weng Marc, & Pattnaik, Debidutta. (2021). Artificial intelligence and machine learning in finance: Identifying foundations, themes, and research clusters from bibliometric analysis. Journal of Behavioral and Experimental Finance, 32, 100577.

Hassoun, Abdo, Aït-Kaddour, Abderrahmane, Abu-Mahfouz, Adnan M., Rathod, Nikheel Bhojraj, Bader, Farah, Barba, Francisco J., Biancolillo, Alessandra, Cropotova, Janna, Galanakis, Charis M., & Jambrak, Anet Režek. (2022). The fourth industrial revolution in the food industry—Part I: Industry 4.0 technologies. Critical Reviews in Food Science and Nutrition, 1–17.

Kavitha, M., Gnaneswar, G., Dinesh, R., Sai, Y. Rohith, & Suraj, R. Sai. (2021). Heart disease prediction using hybrid machine learning model. 2021 6th International Conference on Inventive Computation Technologies (ICICT), 1329–1333. IEEE.

Kotabe, Masaaki Mike, & Helsen, Kristiaan. (2022). Global marketing management. John Wiley & Sons.

Machkour, Badr, & Abriane, Ahmed. (2020). Industry 4.0 and its Implications for the Financial Sector. Procedia Computer Science, 177, 496–502.

Madakam, Somayya, Holmukhe, Rajesh M., & Jaiswal, Durgesh Kumar. (2019). The future digital work force: robotic process automation (RPA). JISTEM-Journal of Information Systems and Technology Management, 16.

Munirathinam, Sathyan. (2020). Industry 4.0: Industrial internet of things (IIOT). In Advances in computers (Vol. 117, pp. 129–164). Elsevier.

Rahmayati, Rahmayati. (2021). Competition Strategy In The Islamic Banking Industry: An Empirical Review. International Journal Of Business, Economics, And Social Development, 2(2), 65–71.

Sun, Huidong, Rabbani, Mustafa Raza, Sial, Muhammad Safdar, Yu, Siming, Filipe, José António, & Cherian, Jacob. (2020). Identifying big data’s opportunities, challenges, and implications in finance. Mathematics, 8(10), 1738.

Wójcik, Dariusz, & Ioannou, Stefanos. (2020). COVID‐19 and finance: market developments so far and potential impacts on the financial sector and centres. Tijdschrift Voor Economische En Sociale Geografie, 111(3), 387–400.

Xie, Xiufeng, Zhang, Xinyu, & Zhu, Shilin. (2017). Accelerating mobile web loading using cellular link information. Proceedings of the 15th Annual International Conference on Mobile Systems, Applications, and Services, 427–439.