Nexus of Intelligent Financial Inclusion and Sustainability: Leveraging Fintech Innovations to Overcome Barriers to Financial Inclusion

A Conceptual Matrix of Barriers to Financial Inclusion

A barrier to financial inclusion is defined as any obstacle that limits or prevents individuals from accessing basic and formal financial services. Financial exclusion in an economy can be voluntary or involuntary. Literature mentions that factors of access to financial products are from supply-side and usage involves demand-side factors (Camara & Tuesta, 2014). Voluntary financial exclusion is caused by the will of people from not using basic financial services (Kebede et al., 2021; Martínez et al., 2013). It may include lack of funds, high costs, or distance of branches (Ezzahid & Elouaourti, 2021). These barriers represent behaviors of people, non-affordability or unawareness (Beck et al., 2006). On the other side, involuntary financial exclusion is not caused by personal choices, rather this is caused by the inability of service providers (Allen et al., 2016; Johnson & Arnold, 2012; World Bank, 2014). Incapacity or failure of financial service providers represent significant supply-side barriers. Expensive financial services, information barrier, non-customized products (Beck & De La Torre, 2007; Honohan, 2005), less lending facilities to poor people due to high risk (Fischer, 2011; IFC, 2015) are significant supply-side barriers.

A review of literature highlights that there are not clear boundaries among demand-side and supply-side barriers. Some of the significant barriers are also not mentioned in the literature. For instance, the economic system of a country has a huge impact on financial sector like political situations or currency risk. In this regard, this research paper introduces “system barriers” as important obstacle to financial inclusion. These are defined as barriers within the overall economic system of a country that have a significant impact on financial inclusion. Other factors like distance, unaffordability, documentation requirements are the perceptions of people that they are unable to use financial services. This discussion implies the need for clear distinctions between voluntary and involuntary financial exclusion, demand-side and supply-side barriers, usage and access barriers, as well as perceived and systemic barriers. To address this gap in literature and for the provision of comprehensive understanding of all barriers to financial inclusion, this research presents a matrix for clear understanding of all dimensions. This contribution is significant with respect to effective solutions that can only be proposed once the problems are precisely identified. Figure 1 represents this matrix.

OPEN IN VIEWER

Figure 1.

A conceptual matrix of barriers to financial inclusion. Source: Authors’ description

Another objective of this research article is to provide support for intelligent financial inclusion. Researchers have selected barriers based upon the importance and their solution in intelligent system. Yasir et al. (2022) identified 16 barriers to financial inclusion which have a direct solution in AI. After analyzing the identified barriers, several were either eliminated or consolidated for analysis based on their overlapping characteristics. For instance, transactional and operational costs were merged under the broader category of high costs due to their inherent similarities. Following the modifications, a total of 10 barriers were finalized for analytical purposes. These barriers serve as key parameters in the present study.

Understanding Barriers to Financial Inclusion: Basis for Fuzzy Inference Mechanism

The Role of AI in Mitigating Barriers to Financial Inclusion

AI has changed the operations of financial industry dramatically (Ryll et al., 2020). Specifically, the role of AI is mitigating barriers to financial inclusion and enhancing its levels is mentioned in the literature (Fazal et al., 2025; Kshetri, 2021; Senyo & Osabutey, 2020; Yasir et al., 2022). A renowned name in the literature is of Mhlanga (2020) who used documentary analysis to provide evidence on the role of AI on financial inclusion. Some of the supporting examples include dramatic impact of M-Pesa, digitalization of China’s financial system and increase in financial access in India as a result of “India stack” (AFI, 2018). Literature mentions that any kind of fintech innovation plays a positive role in enhancement of financial inclusion in the society (Barruetabeña, 2020; Elia et al., 2022; Hua et al., 2019). Ease of opening accounts, attracting a large number of consumers, robo-advisory services, AI credit scoring models and changing trends of technology have shifted the society towards more involvement of people in financial sector (Hentzen et al., 2021; Lee, 2021; Shanmuganathan, 2020). ML techniques use big data for building credit scores of customers in lieu of traditional data (Agarwal et al., 2020; Bazarbash, 2019; Óskarsdóttir et al., 2020). This approach works for the improvement of credit scoring and assist in the involvement of more people especially youngsters and old ones in terms of access to formal financial credit. Fazal et al. (2023) presented systematic literature review on the topic of AI and financial inclusion. This paper has provided evidence on the strong role of AI and ML in enhancing the level of financial inclusion in an economy.

Below Table 1 provides a summary of the role of AI in mitigating different barriers to financial inclusion. The classification of barriers is taken from the matrix presented in the previous section.

OPEN IN VIEWER

Table 1.

Description of AI in response to barriers to financial inclusion

Barriers to financial inclusion	AI as a solution to barriers to financial inclusion
Demand-side_ Usage barriers	The introduction of AI applications on smartphones informs consumers about their financial choices. For instance, the instructional app TymeCoach from TymeBank helps people manage their finances (Demirgüç-Kunt et al., 2017; Schuetz & Venkatesh, 2019;). AI programs can also recognize the religious commitments of people and help in their compliance. ALGEBRA in Malaysia and BIMB investment are two renowned examples in this regard (Kshetri, 2021).
Supply-side_ Access barriers	Techniques for big data and cloud computing can collect information from various sources and disseminate it electronically to everyone. Greeman Bank of Bangladesh is a well-known institution that provides loans to disadvantaged groups in society (Mhlanga, 2020; Wang & He, 2020). Algorithmic trading in stock markets deal with huge amounts of data efficiently and a lot of transactions are performed in milliseconds (Buchanan, 2019; Donepudi, 2019;). With the help of AI and ML, loan applications are processed in minutes. Mexico’s Kueski, which provide loans to clients ineligible for conventional financial services, is a good example. Another program that helps Americans with their daily money issues is Clerkie. (Demirguc-Kunt, Klapper, Singer, Ansar, & Hess, 2017; Kshetri, 2021; Schuetz & Venkatesh, 2019;). AI can review 12,000 documents in a matter of seconds, whereas humans need about 360,000 h to complete the same task. In a single day at a 2017 shopping event, Ant Group chatbot AliMe answered 9 million questions. Customers in Canada now just have to wait 2.5 min instead of an hour thanks to Telus’ chatbot. TymeBank, a digital retail bank without a physical branch in South Africa, saves consumers time by allowing them to open accounts in as little as 5 min. AI algorithms, KYC, and AMLL techniques are being used to apply AI to office operations at all levels (Buchanan, 2019; Decosmo, 2019; Donepudi, 2019;). Robo-advisory platforms, such as Halofina in Indonesia and ALGEBRA in Malaysia, offer investment solutions that align with the objectives of its clients. By making collection recommendations based on a customer’s profile, AI and ML algorithms lower the delinquency ratio. To expand its collections, the Colektia company in Chile uses AI technology (Klapper & Singer, 2017).
Demand side_ Perceived barriers	Fintech’s use in everyday operating scenarios contributes to customers’ increased trust in the system. Furthermore, AI’s ability to adapt and transform is exceptional. Customers’ fears will be allayed when AI helps with savings and investment decisions (Kshetri, 2021). Manual fund transfers require more than 150 keystrokes and take eight to 10 min. It takes less than a minute for such an exchange to be automated by robotics (Mancher et al., 2017). Because financial services like insurance are expensive, people do not choose them. The advent of technology lowers these expenses. According to Tulasi et al. (2017), technological advancements have significantly reduced the fixed transactional costs of financial products in India.
Supply-side_ System barriers	With the revolution in technology, master cards are manufactured with AI chips to alerts owners about usage by fraudsters. One example is Pagshield in Brazilian market that is AI-based anti-fraud tool (Al-Blooshi & Nobanee, 2020; Donepudi, 2019; Perez & Soo, 2017;). In Indonesia, Crowdo platform analyzes the risk of the business against uncertain events, resulting in only 2% of non-performing loans. Same case is Ceptinel company in Chile that alerts for fraud and malpractices by using data from different sources (IMF, 2021). The rise of Distributed Ledger Technologies (DLTs) have provided the platform for secure transactions. Moreover, digital currencies including cryptocurrencies and Central Bank Digital Currencies (CBDCs) are covering the risks of traditional currencies and exchange rates (Mhlanga, 2020). Financial regulations are quite important for the smooth functioning of financial sector in any economy. Jayasekara (2021) who created an index of AML/CFT compliance for ranking countries, has demonstrated this fact in their research.

Intelligent Financial Inclusion: A Way Forward

The proposal of this research article is to implement intelligent systems in the financial infrastructure of any country. Intelligence is a human ability that signifies learning and thinking. Intelligence science is an interdisciplinary field taking concepts from brain science, cognitive science, and artificial intelligence (Shi, 2009). Specifically, the term intelligence in the mimicry of human intelligence in machines, that is, AI. Russell and Norvig (2003) mentioned four modes of AI: Acting humanly, thinking humanly, thinking rationally, and acting rationally. Antsaklis (1999) explained intelligent control as a field where control mechanisms are able to learn, plan and change as the characteristics of human intelligence (Meyer et al., 2009). These control modes work on the basics of computer operations, mathematics, and biological sciences. These systems serve as a base for robotics, communications, fuzzy, expert and hybrid systems.

Intelligent financial inclusion is the implementation of characteristics of the human mind, cognitive sciences, and AI tools and applications in the financial sector to provide efficient and suitable financial products and services to all the segments of population by removing the barriers to the traditional financial system. This system forecasts solutions and their applicability for clients by learning from the data and parameters that are already available. The only way to overcome financial exclusion and achieve the ideal degree of financial inclusion is through such a system. The ultimate outcome of this IFI will be the effective and reasonably priced provision of financial services and products to all societal sectors, together with a decrease in fraud and malpractice and the provision of appropriate services to the appropriate individuals and locations. Scholarly work is demonstrating this fact that financial inclusion and sustainable finance is associated with reduction in hazardous effects of climate including its role in decarbonization (Said & Acheampong, 2024; Zhang et al., 2026). Research model of this research takes the support from systems theory of financial inclusion (Kim et al., 2018; Ozili, 2018). It states that financial inclusion relies on all the sub-systems of economy including social and economic systems also, which in return affects the provision of financial services in an economy. Systems theory addresses the main purpose of this research by involving all the people in financial sector. Figure 2 represents the barriers selected for analysis in current research with their dimensions, to achieve intelligent financial inclusion by removing these barriers from economic system.

OPEN IN VIEWER

Figure 2.

Research model, Layer 1 consists of barriers, while Layer 2 presents their dimensions

Fuzzy Inference Mechanism

For the current study, the researchers adopted pragmatic paradigm that integrates interpretivist insights with quantitative reasoning. At first, they have explored the literature subjectively to provide inputs for the fuzzy inference mechanism. Parameters that are barriers to financial inclusion have been taken from a deep review of literature in line with interpretivism. This research is based upon theoretical and conceptual simulation of fuzzy logic using ranges derived from literature review. The defined parameters represent generalized linguistic scales that provide conceptual understanding of the intensity of variables rather than exact numerical values. Within this framework, fuzzy logic serves as a bridge between qualitative perceptions and quantitative analysis by translating them into measurable linguistic variables. Fuzzy logic is an ML tool that supports loose end data having unclear boundaries. The idea of fuzzy logic was presented by Zadeh (1965) who worked on Memberships Functions (MFs) to develop fuzzy logic as a qualitative tool for analyzing hypothesis (Jahantigh, 2019). Fuzzy logic has been widely applied across diverse research domains as an effective approach for modeling uncertainty and supporting complex decision-making and analytical processes (Estefania-Salazar et al., 2025; Wurster & Hagemann, 2020).

In this research, Mamdani fuzzy logic has been used. Fuzzification, execution, and defuzzification are its pillars (Mamdani, 1974). Based on MFs, fuzzification assigns a numerical value to the system input. Uncertainty regarding a situation is represented by any number between 0 and 1. A value of 0 indicates that it has no relationship to fuzzy sets, whereas a value of 1 indicates that it belongs to a fuzzy system. The foundation of the inferencing process is fuzzification, from which data is sent to processing control systems (Almeida et al., 2017). After fuzzification, rules are formulated to generate output functions. At the end, defuzzification generates output based on input sets and corresponding MFs. Figure 3 represents the fuzzy inference model of current study.

OPEN IN VIEWER

Figure 3.

Model of fuzzy inference mechanism crisp input → fuzzified values → fuzzy inference → crisp output

Input Variables

The fuzzy logic framework proposed in this study utilizes the statistical values of 10 input parameters (barriers to financial inclusion) to evaluate whether a system reflects a state of intelligent financial inclusion or otherwise. The classification scale is derived from an extensive review of the literature, specifically considering the varying intensities with which barriers impact financial inclusion. Each level on this scale is assigned a corresponding numerical value, representing the relative strength of each barrier’s influence on the outcome. These quantified input values are integral to generate the system’s results in a mathematical form. Table 2 outlines the scales, numerical values, and MFs of the Layer-1 and Layer-2 input parameters.

OPEN IN VIEWER

Table 2.

Assignment of scale, numerical values and MFs of both layers used in the analysis

Layer-1				Layer-2
Barrier	Scale/linguistic variable	Membership functions		Sub-factors of barrier (20)	Scale/linguistic variable	Membership functions
Information asymmetry	Low Medium High	0–25 25–50 50–100	[0 0 15 25] [15 25 40 50] [40 50 100 100]	1. Structural barriers	Low	0–30	[0 0 20 30]
					Medium	30–60	[20 30 50 60]
					High	60–100	[50 60 100 100]
				2. Strategic & behavioral barriers	Low	0–30	[0 0 20 30]
					Medium	30–60	[20 30 50 60]
					High	60–100	[50 60 100 100]
Inefficient processes	Less Average More	0–30 30–60 60–100	[0 0 15 30] [15 30 45 60] [45 60 100 100]	1. Manual systems	Less	0–30	[0 0 15 30]
					Average	30–60	[15 30 45 60]
					More	60–100	[45 60 100 100]
				2. Time consuming	Less	0–25	[0 0 15 25]
					Average	25–50	[15 25 40 50]
					More	50–100	[40 50 100 100]
				3. High error rate	Low	0–15	[0 0 10 15]
					Average	15–30	[10 15 25 30]
					High	30–100	[25 30 100 100]
High costs	Low Normal High	0–20 20–40 40–100	[0 0 10 20] [10 20 30 40] [30 40 100 100]	1. High transactional costs	Low	0–20	[0 0 10 20]
					Normal	20–40	[10 20 30 40]
					High	40–100	[30 40 100 100]
				2. High operational costs	Low	0–20	[0 0 10 20]
					Normal	20–40	[10 20 30 40]
					High	40–100	[30 40 100 100]
Poor credit risk analysis	Low Normal High	0–30 30–60 60–100	[0 0 20 30] [20 30 50 60] [50 60 100 100]	1. Manual operations	Less	0–30	[0 0 15 30]
					Normal	30–60	[15 30 45 60]
					High	60–100	[45 60 100 100]
				2. Time consumption	Less	0–30	[0 0 20 30]
					Average	30–60	[20 30 50 60]
					More	60–100	[50 60 100 100]
				3. Biasness	Less	0–25	[0 0 15 25]
					Average	25–50	[15 25 40 50]
					More	50–100	[40 50 100 100]
				4. High costs	Low	0–30	[0 0 15 30]
					Normal	30–60	[15 30 45 60]
					High	60–100	[45 60 100 100]
Risk	Low Normal High	0–30 30–60 60–100	[0 0 10 30] [10 30 40 60] [40 60 100 100]	1. Country risk	Low	0–30	[0 0 10 30]
					Normal	30–60	[10 30 40 60]
					High	60–100	[40 60 100 100]
				2. Market risk	Low	0–30	[0 0 10 30]
					Normal	30–60	[10 30 40 60]
					High	60–100	[40 60 100 100]
				3. Currency risk	Low	0–30	[0 0 10 30]
					Normal	30–60	[10 30 40 60]
					High	60–100	[40 60 100 100]
				4. Fraud	Low	0–30	[0 0 10 30]
					Normal	30–60	[10 30 40 60]
					High	60–100	[40 60 100 100]
Regulatory issues	Few Average High	0–30 30–60 60–100	[0 0 15 30] [15 30 45 60] [45 60 100 100]	1. Rigid regulations	Less	0–40	[0 0 20 40]
					Average	40–80	[20 40 60 80]
					More	80–100	[60 80 100 100]
				2. Challenge for innovation	Low	0–25	[0 0 15 25]
					Average	25–50	[15 25 40 50]
					High	50–100	[40 50 100 100]
Financial illiteracy	Low Average High	0–30 30–60 60–100	[0 0 15 30] [15 30 45 60] [45 60 100 100]	1. Poor financial knowledge	Poor	0–30	[0 0 10 30]
					Average	30–60	[10 30 40 60]
					Good	60–100	[40 60 100 100]
				2. Bad financial behavior	Bad	0–40	[0 0 20 40]
					Normal	40–80	[20 40 60 80]
					Good	80–100	[60 80 100 100]
				3. Negative financial attitude	Negative	0–30	[0 0 20 30]
					Normal	30–60	[20 30 50 60]
					Positive	60–100	[50 60 100 100]
Inefficient customer service	Less Average More	0–30 30–60 60–100	[0 0 15 30] [15 30 45 60] [45 60 100 100]	​	​	​	​
Dishonest advisory service	Low Medium High	0–30 30–60 60–100	[0 0 10 30] [10 30 40 60] [40 60 100 100]	​	​	​	​
Heavy documentation	Low Normal High	0–20 20–40 40–100	[0 0 10 20] [10 20 30 40] [30 40 100 100]	​	​	​	​

Defuzzification

Fuzzy application generates output from respective inputs by defuzzification. Output parameters acquire results corresponding to input parameters. The proposed system of this research article has three output states. A country can have low, average and high levels of intelligent financial inclusion when AI is being implemented in financial infrastructure to remove barriers to financial inclusion (Figure 4). In MATLAB, intelligent financial inclusion is represented as IFI.

OPEN IN VIEWER

Figure 4.

MFs of output “Intelligent Financial Inclusion”

Simulation Results

There are two input variables in the demand-side barriers to financial inclusion and eight in supply-side barriers in current fuzzy inference mechanism Lookup tables map the combinations of fuzzy input sets to corresponding outputs based on fuzzy rules. One lookup table from layer one is shown in Figure 5. The values represent that when financial illiteracy and heavy documentation has high values, IFI is low. IFI generates a value of 15.3 when financial illiteracy is 67.4 and heavy documentation is equal to 65.

OPEN IN VIEWER

Figure 5.

Look up diagram of proposed IFI by demand-side variables (financial illiteracy and heavy documentation)

Layer 2 consists of 20 sub-factors of 10 barriers to financial inclusion. Let’s consider the breakup of second parameter, that is, inefficient processes. There are three sub-factors of inefficient processes; therefore, 27 rules were formed. Like, if (manual system is less) and (time-consuming is average) and (high error rate is average) then (inefficient processes is average). According to lookup table, high values of time-consuming processes and high error rates are cause of more inefficient processes even though the manual system has some average value Figure 6. Rest of lookup tables are presented in Appendix 3.

OPEN IN VIEWER

Figure 6.

Look up diagram of inefficient processes for manual system, time-consuming and high error rates

Defuzzification

Defuzzification is an important element of fuzzy inference mechanism which generates output and highlights areas for operations. Figure 7 is the rule surface of IFI by demand-side input parameters of financial illiteracy and heavy documentation. Yellow area is excellent IFI with minimum intensity of financial illiteracy and heavy documentation. Greenish area is satisfactory and average while blue area represents the poor results for IFI with maximum intensity of financial illiteracy and heavy documentation.

OPEN IN VIEWER

Figure 7.

Rule surface of IFI by demand-side input parameters of financial illiteracy and heavy documentation (Layer 1)

A total of eight supply-side input parameters were considered for the analysis of IFI. Representative defuzzification graphs are presented in Figures 8 and 9. Figure 8 illustrates the fuzzy inference system for IFI with inefficient processes and high costs as inputs. The 3D surface plot reveals that lower values for both inputs correspond to higher IFI scores. The yellow region denotes excellent performance (low input intensity, high output), green indicates satisfactory performance, and blue reflects deficient performance (high input intensity, low output). An inverse relationship between inefficient processes and high costs is shown in Figure 9. IFI reaches its maximum when both inputs are at zero (yellow) and declines to zero when inputs reach 100 (dark blue).

OPEN IN VIEWER

Figure 8.

Rule surface of IFI using inefficient processes and high costs

OPEN IN VIEWER

Figure 9.

Rule surface of IFI using high costs and inefficient processes

This study utilizes 10 parameters as input variables in a fuzzy logic-based framework to assess Intelligent Financial Inclusion (IFI). Each input is analyzed through its sub-factors to enhance clarity, for instance, financial illiteracy is broken down into poor financial knowledge, behavior, and attitude. Most variables are further modeled in layer 2, except for heavy documentation and dishonest advisory services. Multiple 3D surface graphs in defuzzification illustrate how intensities of inputs influence outputs. For example, high structural and behavioral barriers increase information asymmetry, which negatively affects IFI (Figure 10). Financial illiteracy rises with stronger negative financial behaviors and attitudes (Figure 11). Each relationship is visually represented through 3D graphs, demonstrating the varying degrees of influence input intensities exert on their respective outcomes.

OPEN IN VIEWER

Figure 10.

Rule surface of information asymmetry based upon structural and strategic & behavioral barriers

OPEN IN VIEWER

Figure 11.

Rule surface of financial illiteracy based upon poor financial knowledge and bad financial behavior

There are certain practical examples of use of AI and other innovative technologies in financial sector that has increased the level of financial inclusion among economies. Greeman bank of Bangladesh has used innovative technologies to automate their processes in reaching underprivileged segments of population that increased the involvement of people in financial sector. One of the successful example is Mexico’s Kueski that reduced the transactional costs of banks to almost 0. An online form in filled in for 2 min and the amount is credited in account within 8 min (Fazal et al., 2025). Another example is AI-based equities fund introduced by BIMB Investment in Malaysia that helped to comply with Shariah, religious, and regulatory criteria. Such examples demonstrate that all the barriers are influential in reducing the level of financial inclusion in economies.