Toward an evolving framework for responsible AI for credit scoring in the banking industry
| Date | 07 January 2025 |
| Pages | 148-163 |
| DOI | https://doi.org/10.1108/JICES-08-2024-0122 |
| Published date | 07 January 2025 |
| Author | Manoj Philip Mathen,Anindita Paul |
Toward an evolving framework
for responsible AI for credit scoring
in the banking industry
Manoj Philip Mathen and Anindita Paul
Indian Institute of Management Kozhikode, Kozhikode, India
Abstract
Purpose –The aim of this research is to conductasystematic review of the literatureon responsible artificial
intelligence (RAI) practices within thedomain of AI-based Credit Scoring (AICS)in banking. This review
endeavours to map the existing landscapeby identifying the work done so far, delineating the key themes and
identifyingthe focal points of research within this field.
Design/methodology/approach –A database search of Scopusand Web of Science (last 20 years)resulted
in 377 articles. Thiswas further filtered for ABDC listing, and augmented withmanual search. This resulted in
afinal list of 53 articles which was investigatedfurther using the TCCM (Theory, Context, Characteristicsand
Methodology)review protocol.
Findings –The RAI landscape for credit scoring in the banking industry is multifaceted, encompassing
ethical, operational and technological dimensions. The use of artificial intelligence (AI) in banking is
widespread, aiming to enhance efficiency and improve customer experience. Based on the findings of the
systematic literaturereview we found that past studies on AICS have revolved around four major themes:(a)
AdvancesinAItechnology; (b) Ethical considerations and fairness;(c) Operational challenges and limitations;
and (d) Future directions and potential applications. The authors further propose future directions in RAI in
credit scoring.
Originality/value –Earlier studies have focused on AI in banking,credit scoring in isolation. This review
attemptstoprovidedeeper insights, facilitating the developmentof this key field.
Keywords Credit scoring, Responsible AI, Ethics, Fairness, Explainable AI, Algorithms, AI
Paper type Literature review
Introduction
Credit scoring is a fundamental processused by banks and financial institutions to assess the
creditworthiness of individualsand businesses. Initially introduced by Altman (1968), credit
scoring has evolved significantly, with modern methodologies leveraging sophisticated
models to support decision-making in loan approvals (Aggarwal, 2021;Yang, 2007).
Traditionally, credit scoring involved analysing factors such as credit history, income and
debt-to-income ratio (Onay and Öztürk, 2018;Talaat et al., 2024). However, the advent of
FinTech and AI-driven systems has revolutionized this field, introducing advanced
techniques like machine learning (ML) and pattern recognition (Langenbucher and
Corcoran, 2021;MarquésA I et al., 2013).
The integration of AI (artificial Intelligence)in credit scoring, particularly through theuse
of AI Credit Scoring Systems (AICS), offers unprecedented accuracy in predicting
creditworthiness by analysing vast amounts of data (Buyl and Bie, 2024;Giudici, 2018).
Despite these advancements, the black-box nature of these systems raises significant ethical
concerns, particularly regarding transparency, fairness and accountability (Kear, 2017;
Kumar and Suthar, 2024). Responsible AI responsible artificial intelligence (RAI) seeks to
JICES
23,1
148
Received27 August 2024
Revised18December2024
Accepted19 December2024
Journalof Information,
Communicationand Ethics in
Society
Vol.23 No. 1, 2025
pp. 148-163
© Emerald Publishing Limited
1477-996X
DOI 10.1108/JICES-08-2024-0122
The current issue and full text archive of this journal is available on Emerald Insight at:
https://www.emerald.com/insight/1477-996X.htm
address these issues by ensuringthat AI systems are developed and used in ways that are fair,
transparent and accountable,thereby mitigating potential biases and discrimination(Mikalef
et al., 2022). The importance of RAI in creditscoring is underscored by the growing reliance
on AI in this domain, which introduces new challenges,such as ensuring that AI algorithms
do not perpetuate or amplify existing biases in the data.
In recent years, the use of non-traditional data sources, such as social mediaactivity and
psychometric data, has further complicatedthe landscape of credit scoring, introducing new
challenges and opportunities for banks and FinTech companies (Kallus et al., 2022;
Campbell-Verduyn et al., 2017). The increasing reliance on AI in this domain underscores
the need for a systematic review of the literature on Responsible AI in Credit Scoring
(RAICS), particularly in light of concerns about potential biases and ethical dilemmas
(Anagnostou et al.,2022;Jobin et al., 2019). This review identifies the key themes and gaps
in the current literature that also serves as a foundational step towards the development of
comprehensive frameworksfor responsible AI adoption in credit scoring.
Conducting a systematic literature review (SLR) is crucial in this context, as Paul et al.
(2021) highlight its importance when a field is rapidly evolving, the literature is expanding
and there is a need to synthesize diverse findings. With AI’s rapid advancements in credit
scoring and the associated ethical challenges, this study aims to systematically review the
literature on RAICS within the banking industry. By using the TCCM (Theory, Context,
Characteristics and Methodology) framework (Paul and Rosado-Serrano, 2019), we will
organize and synthesize existing research, identify gaps and propose directions for future
studies. To the best of our knowledge, this study is the first attempt toward establishing a
framework on RAICS in banking, offeringa comprehensive analysis of its impact on ethical
decision-makingin credit scoring.
The structure of this paper is as follows: Section 2 outlines the research methodology;
Section 3 discusses the current publication trends; Section 4 presents the key findings of the
literature review and Section 5 providesimplications for research and practice, along with
suggestions for futurestudies.
Research methodology
SLRs synthesize existing knowledge in a systematic, transparent manner to guide future
research priorities (Page et al.,2021;Davis et al.,2014). SLRs address questions beyond the
scope of individual studies, making them essential for deve loping emerging fields like RAICS.
In the following sections we explain our criteria for shortlisting the papers reviewed in this SLR.
The search strategy involved a two-step process: an electronic database search and a manual
search of key journals. The databases Scopus and Web of Science were selected for their
comprehensive coverage of high-impact journals. Keywords were carefully chosen based on the
key constructs of the study –banking, credit scoring and Responsible AI (Figure 1).
For the electronic search, the initial keywords included “AI,”“Responsible AI,”and “Credit
Scoring.”The keywords listed in the “keywords”section of the identified articles were reviewed
to refine the search. Additional relevant terms were identifiedbasedonrecurringthemesand
constructs within the literature. The final set of keywords were –AI, Artificial Intelligence,
Algorithm, Credit Scoring, Credit Assessment, Lending, Ethics, Bias, Fairness, Responsible,
Explainable AI, Trust, Discrimination and Regulation. The final keywords were combined
using Boolean operators “AND”and “OR”to conduct a comprehensive search across Scopus
and Web of Science databases. The results were then filtered by reading the abstracts and
keywords to ensure relevance. The search string was: RAI OR AI OR Artificial Intelligence OR
Algorithm) AND (Credit Scoring OR Credit Assessment or Lending) AND (Ethics OR Bias
Journal of
Information,
Communication
and Ethics in
Society
149
Get this document and AI-powered insights with a free trial of vLex and Vincent AI
Get Started for FreeUnlock full access with a free 7-day trial
Transform your legal research with vLex
-
Complete access to the largest collection of common law case law on one platform
-
Generate AI case summaries that instantly highlight key legal issues
-
Advanced search capabilities with precise filtering and sorting options
-
Comprehensive legal content with documents across 100+ jurisdictions
-
Trusted by 2 million professionals including top global firms
-
Access AI-Powered Research with Vincent AI: Natural language queries with verified citations
Unlock full access with a free 7-day trial
Transform your legal research with vLex
-
Complete access to the largest collection of common law case law on one platform
-
Generate AI case summaries that instantly highlight key legal issues
-
Advanced search capabilities with precise filtering and sorting options
-
Comprehensive legal content with documents across 100+ jurisdictions
-
Trusted by 2 million professionals including top global firms
-
Access AI-Powered Research with Vincent AI: Natural language queries with verified citations
Unlock full access with a free 7-day trial
Transform your legal research with vLex
-
Complete access to the largest collection of common law case law on one platform
-
Generate AI case summaries that instantly highlight key legal issues
-
Advanced search capabilities with precise filtering and sorting options
-
Comprehensive legal content with documents across 100+ jurisdictions
-
Trusted by 2 million professionals including top global firms
-
Access AI-Powered Research with Vincent AI: Natural language queries with verified citations
Unlock full access with a free 7-day trial
Transform your legal research with vLex
-
Complete access to the largest collection of common law case law on one platform
-
Generate AI case summaries that instantly highlight key legal issues
-
Advanced search capabilities with precise filtering and sorting options
-
Comprehensive legal content with documents across 100+ jurisdictions
-
Trusted by 2 million professionals including top global firms
-
Access AI-Powered Research with Vincent AI: Natural language queries with verified citations
Unlock full access with a free 7-day trial
Transform your legal research with vLex
-
Complete access to the largest collection of common law case law on one platform
-
Generate AI case summaries that instantly highlight key legal issues
-
Advanced search capabilities with precise filtering and sorting options
-
Comprehensive legal content with documents across 100+ jurisdictions
-
Trusted by 2 million professionals including top global firms
-
Access AI-Powered Research with Vincent AI: Natural language queries with verified citations