Artificial intelligence (AI) will fundamentally change the financial sector for years to come. The integration of AI technologies – such as machine learning, natural language processing, generative AI (GenAI), and advanced data analytics – will enable financial institutions to optimize their operations, enhance customer experiences, develop innovative products, and better manage risks. AI is already reshaping how individuals access and manage financial services, from real-time fraud detection to credit decisions based on mobile phone data, and is poised to accelerate these changes rapidly.
According to McKinsey, the value generation potential of GenAI in the banking sector resulting from productivity gains is estimated at 200-340 billion USD in increased annual revenue, equivalent to 2.8-4.7% of total industry revenue globally, without counting gains resulting from better risk management. In view of AI’s large potential, the financial sector’s global spending on AI is projected to increase from 35 billion USD in 2023 to 126 billion USD in 2028. Some forecasts suggest AI could enable financial institutions to process transactions 90% faster and potentially allow the sector to save 1 trillion USD globally by 2030.
Leveraged properly, this AI revolution holds immense promise for financial inclusion.
Leveraged properly, this AI revolution holds immense promise for financial inclusion. But this promise cannot happen alone and will not happen automatically. The financial inclusion sector needs to invest in developing new models of banking that leverage this opportunity to serve low-income segments. The potential impact is likely to be both significant and swift.
Embracing AI can create opportunities for low-income customers and providers alike
AI is likely to be the transformative enabler that is needed to design products that better meet more customers’ needs while simultaneously incentivizing financial service providers (FSPs) to serve traditionally underserved customers at the last mile. It can solve major barriers to financial inclusion by reducing costs, better tailoring products, bridging information gaps, and building trust.
Reducing costs
AI can substantially reduce costs throughout the financial service value chain, fundamentally altering the cost of customer acquisition and retention and of processing financial transactions – which in turn is likely to make it increasingly worthwhile for FSPs to serve low-income customers despite their smaller-value transactions.
Over the last few years, the cost for a financial institution of acquiring new clients has decreased significantly with digitization. For example, India’s digital infrastructure, which includes the United Payments Interface (UPI, India’s instant payment system) and Aadhaar (India’s biometric ID system), has reduced the cost of client acquisition for financial institutions from USD 12 to 6 cents, according to the IMF. AI has the power to reduce the costs of acquiring new clients even further by streamlining client due diligence and authentication processes, such as with the launch of India’s Central KYC Registry which uses matching algorithms and facial recognition. Of course, this shows that having a robust digital infrastructure and enabling policies in place is essential to allow countries to capitalize on AI for further efficiencies in the sector.
In addition, by facilitating the automation of processes and the analysis of large datasets, AI is also decreasing the operating costs of financial institutions, through more efficient handling of customer service inquiries, complaint processing, fraud management, and many other back office and operational processes. AI can also assist in conducting better client segmentation analyses, and therefore in better targeting and cross-selling by determining which products to offer to specific customer segments, enhancing product explanations, and identifying features that better suit customers. In addition, AI-powered automated underwriting also enables faster and lower-cost loan approvals. As a result, AI is essentially lowering the cost of handling small-value transactions, making it more profitable and attractive for FSPs to serve lower-income individuals.
Better tailoring of products
Traditionally, financial institutions have provided standardized one-size-fits-all products that do not always meet the needs of the underserved, leaving many without access to essential financial services. AI can reform FSPs’ ability to collect data on low-income clients by relying on digital data trails and/or big data instead of costly surveys. Additionally, AI can facilitate a more cost-effective way to analyze the collected data, thereby enabling a fundamentally new approach to design for inclusion, rooted in deep, real-time insights into the needs and cash flows of low-income clients. This is already enabling the design of tailored products at low cost.
For example, fintechs like Boost in Nigeria, Copia in Kenya, Fairbanc in Indonesia, and Mercado Pago in Brazil, have developed embedded finance solutions that integrate last-mile retailers onto digital ordering platforms, allowing them to access tailored inventory financing on reasonable terms based on detailed analyses of their cash flows. AI-informed embedded finance and tailored loans can go a long way in reducing the 5 trillion USD finance gap for micro, small, and medium enterprises (MSMEs).
Applying AI-driven conversational interfaces can also be a game changer for improving the suitability of financial products for individuals in remote areas and those with low literacy. AI voice assistants and chatbots can enable customers to interact with financial institutions 24/7 in their own language while guiding them through complex financial processes via natural language processing explanations. This has the potential to fundamentally change the customer’s experience in accessing financial services, by overcoming barriers of language, literacy, and location.
Applying AI-driven conversational interfaces can also be a game changer for improving the suitability of financial products for individuals in remote areas and those with low literacy.
Furthermore, AI can enable the joint provision of financial and non-financial services, such as providing weather information for smallholder farmers, or investment and business advice for micro-entrepreneurs, which enhances the impact of financial services. For instance, Digital Green developed Farmer.Chat, an AI-powered chatbot designed to provide real-time, personalized agricultural advice to small-scale farmers alongside financial services.
As AI’s capabilities and accessibility evolve, its potential for financial inclusion becomes even more revolutionary. For instance, AI could analyze diverse datasets to create personalized insurance plans for low-income individuals, tailoring coverage to their specific needs. It could also facilitate the creation of micro-saving accounts that cater to the unique financial habits of underserved populations and could help design pension plans that are commensurate to the savings potential of informal sector workers, who often lack access to traditional pension schemes.
Bridging information gaps
For many years, accessing loans was out of reach for low-income individuals due to their lack of credit history and collateral. However, growing digital trails and AI models have started to shift how FSPs conduct credit scoring and credit risk assessment by moving away from historical credit histories, making low-income customers more viable for loans. Specifically, AI algorithms can use alternative data sources, like digital transactions, utility bill payments, inventory purchases, or sale cash flows, to assess creditworthiness, with high-quality loss predictability.
For instance, Indian fintechs Fundfina (which offers credit to small shops) and KarmaLife (which provides credit for platform workers) have used credit scoring models based on transactional data and achieved similar predictive power to credit history-based models. This can be a game changer to close the very large credit access gaps for both low-income individuals and nano, micro, and small enterprises, which often lack credit histories and collateral but are increasingly generating digital data trails.
Building trust
AI can also enhance customer’s trust in financial institutions by improving fraud detection and prevention, including for small-value transactions. Typically, FSPs have a threshold below which they do not take action on small-value fraud cases or claims, but AI can significantly reduce the cost of fraud management making it possible for FSPs to adequately respond to even low-value cases – thereby contributing to greater customer trust. In addition, AI-based voice assistants and chatbots further create trust by providing more options to assist customers, including real-time assistance in their own language. AI-powered financial literacy apps educate users on savings, investing, and borrowing responsibly, all of which enhance customer confidence in using financial services.
AI can also strengthen the supervision of financial services, for instance by monitoring social media to anticipate consumer risks. By analyzing substantial amounts of data from social media, supervisors can detect early warning signals of consumer harm, such as aggressive debt collection practices or data misuse, and use these findings to shape regulations. For instance, the Central Bank in the Philippines launched a chatbot (BSP Online Buddy) that enables consumers to use SMS or Facebook Messenger to file complaints to the central bank, which then uses SupTech to analyze this data and monitor market misconduct. If scaled responsibly, this kind of AI-enabled supervision could create real-time feedback loops that strengthen consumer protection.
The risks of AI must be proactively managed
Despite AI’s immense potential for financial inclusion, it also brings significant risks, including rapidly increasing fraud, data security breaches, and cyberattacks.
Due to AI’s reliance on massive datasets, its use does inflate the risk of fraud and data breaches through exploiting pattern recognition, enabling the unmasking of anonymized data and other forms of data manipulation. Financial fraud is also increasingly being automated through AI, making it harder to detect and prevent. Low-income customers, who tend to be less digitally and financially savvy than others, are particularly vulnerable.
Additionally, AI creates a risk of enhanced discrimination in financial services due to algorithmic biases. This risk is heightened by the fact that, while data trails are expanding for low-income individuals globally, there is still less data on low-income populations, especially low-income women. These groups remain less digitally connected and public and private databases are not always disaggregated by income groups or gender, which can lead to potential biases in AI models due to insufficient input data on underserved groups. This in turn disproportionately penalizes groups with limited formal data records.
Lastly, without sufficient expertise in AI, financial institutions and supervisors may overly rely on algorithmic outputs without ensuring critical human oversight. Doing so could undermine consumer protection and amplify systemic risks.
Ensuring safe and inclusive AI innovation is imperative
There is no doubt that AI will profoundly transform many aspects of government, business, and society. The same is true of the financial sector and efforts to improve financial inclusion. But to unleash this potential, it is imperative that we invest in AI and leverage its tremendous opportunities for financial inclusion while protecting against its inherent risks.
The choices made now will shape whether AI deepens economic divides or expands opportunity for all.
To start with, investing in connectivity, digital public infrastructure, and inclusive data ecosystems (which would open access to individuals’ data through consent and increase the availability of income- and gender-disaggregated data) will be critical to ensure those who are historically excluded can generate data and benefit from AI-powered financial solutions. In addition, clear and balanced policies, incentives, and supervision are needed to ensure FSPs take advantage of AI solutions to expand their services towards currently excluded or underserved clients, and that AI models employed by FSPs avoid biases and provide fair and positive outcomes for all customers, fostering inclusion, accountability, and transparency. Strengthened consumer protection, data privacy, and cyber security, as well as financial literacy efforts, will also be essential to protect against increased risks and build trust.
New AI models and algorithms are being developed every day, and use cases are being piloted and rolled out even as this essay is being written. This is the window of time to ensure AI improves equality and opportunity, not worsens it. But doing so will require proactive, intentional, and collaborative action.
In other words, the choices made now will shape whether AI deepens economic divides or expands opportunity for all. Will we use it to build a more inclusive and equitable financial system?
This is the first in a new blog series on AI in financial inclusion. Upcoming blogs will delve deeper into the role of the public and private sectors in both addressing the risks associated with AI and also in enhancing the incredible opportunities AI could create for financial inclusion.
