2.1.1. Risk Detection, Management and Measurement

One major reason for many vulnerable groups—like women, youths and small businesses-like smallholder farmers—being excluded from the formal financial market in the traditional banking sector was driven by issues around risk (Beck et al. 2009). Many of these vulnerable groups were viewed as high risk due to the limited capability to detect and measure the risk among them (Park and Mercado 2015, 2018). Some of the factors that exacerbated this was lack of data (Park and Mercado 2018). However, AI is transforming financial inclusion through the widespread use of algorithms to automate risk detection management and measurement (Peric 2015; Muneeza et al. 2018). The use of AI is making it possible for the previously excluded groups to be able to access financial services using various digital tools such as cell phones or instruments like payment cards that can be used to connect with digital devices like point of sale terminals (Alameda 2020; Bill & Melinda Gates Foundation 2019).

In Kenya, M-Pesa, where M represents mobile while Pesa is another word for money in Swahili, is one of the mobile phone-based money transfer service operated by Safaricom which was able to offer payments services, and micro-financing service lunched in 2007 (Osah and Kyobe 2017; Burns 2018). The service has since spread to many countries which include Tanzania, Mozambique, DRC, Lesotho, Ghana, Egypt, Afghanistan, South Africa, India, Romania and Albania among many other countries (Jacob 2016; Burns 2018). The ability of a mobile device using AI intelligence could make it possible for people to make deposits, to withdraw money, to transfer money, pay for goods and services, to have access to credit and savings (Van Hove and Dubus 2019). This helps the low-income earners to be able to access these services which they could not access in the traditional banking system (Wang and He 2020). Additionally, through the use of AI intelligence, registration of accounts was achieved digitally; approximately 17 million accounts were registered in Kenya in its initial stages in 2012 while 7 million accounts were registered in Tanzania in 2016 (Van Hove and Dubus 2019; Wang and He 2020).

AI also plays an important role in preventing currency risk (Paul 2019). Through digital finance, individuals and small businesses (SMEs) have the option to add funds in the fiat currency which allows a shift in the volatility risk to the financial intermediary (FI) (Paul 2019). Many FIs are using bitcoin as a vehicle currency with the United States dollar as the dominant vehicle currency used in 88 per cent of trades (Global Partnership For Financial Inclusion 2016; Paul 2019). The use of bitcoin as a vehicle currency and block chain's platforms means that the recipient and the sender are not exposed to the volatility of virtual currency (Paul 2019). The ability to prevent risk is allowing small income earners to participate in the financial market as a result of the strength of AI technology (Alameda 2020). In short, financial markets are adopting more and more to AI to come with more exciting nimble models which are being utilized by financial experts to pinpoint trends, identify risks, conserve manpower and to ensure better information and for future planning (GPFI 2017).

#### 2.1.2. AI and Information Asymmetry

The credit rationing theory credited to Stiglitz (Berardi 2011). This theory asserts that when information asymmetry (also referred to as imperfect information) is present in a competitive loan market, credit rationing will be the major feature of that credit market. Among a group of borrowers with fully observable and identical characteristics, some will receive loans while others will not get anything (Stiglitz 1989; Yuan et al. 2011). In the process, some disappointed borrowers will be more than willing to pay an interest rate which is more than the market interest rate. However, financial institutions will not be willing to respond to excess demand for loanable funds through raising the interest rate for borrowers (Stiglitz 1989). The major reason given was that in many circumstances when the interest rate is high, safer borrowers do not borrow as they are disuaded from borrowing (Yuan et al. 2011).

In addition, when the interest rate is high, borrowers will invest in high-risk projects which will limit the probability of paying back the loan (Berardi 2011). This condition will limit the participation of other potential players in the credit market. Accordingly, this explanation will help to explain why some economic agents will be excluded in the financial market and the increase in financial exclusion in the formal financial markets. According to the credit rationing theory, one of the major factors which cause the market to malfunction in developing nations is information asymmetry (Bell et al. 1997). It is believed that information asymmetry through adverse selection and moral hazards is the primary source of market inefficiencies (Bell et al. 1997). As a result of these inefficiencies in the market, high-risk borrowers like small scale farmers will be excluded from the group of potential borrowers

(Yuan et al. 2011). This will mark the reason many economic agents are financially excluded in the formal financial markets.

However, digital tools like AI can overcome the problem of information asymmetry (Kaya and Pronobis 2016). Digital financial inclusion through AI can have access to various online shopping platforms and various online social networks which produces a large amount of information on individuals which will help to do away with the problem of information asymmetry between financial institutions and individuals (Wang and He 2020; Yang and Zhang 2020). Digital tools improve access to credit to vulnerable groups especially those without collateral security based on big data analysis and cloud computing (Wang and He 2020). Many digital technologies which use AI technology utilize other credit score mechanisms to create collateral free-loan products (Matsebula and Yu 2017). One example of the bank which offered collateral-free loans was the Grameen Bank that won a Nobel Prize in 2006 together with Prof. Muhammad Yunus. The bank distributed collateral-free loans of united states dollars (USD) 24 billion to borrowers (Karlan and Morduch 2010; Wang and He 2020). In a way, AI solutions are assisting financial institutions and credit lenders to make smarter underwriting decisions through the use of many factors that assess accurately traditionally underserved borrowers in the credit decision-making process (Paul 2019).

#### 2.1.3. AI and Customer Support and Helpdesk through Chatbots

Through the use of AI, banks are now adopting customer support and help desks which are impacting more on increasing efficiency and reducing the cost of customer support. Banks are offering an electronic virtual assistant (EVA). Moreover, with AI, financial institutions can provide personalized banking where chatbots and AI assistants, use AI to come up with personalized financial advice and natural language processing to provide instant, self-help customer service (Alameda 2020; Paul 2019).

Besides, AI is used as a relationship manager, banks are introducing chatbots for this purpose. This allows vulnerable households in rural areas to access financial advice and help which they cannot enjoy when dealing with human beings (Paul 2019). The HDFC bank of India has already introduced a chatbot for relationship manager purposes (Paul 2019). It is alleged that many bank staff have an urban orientation which makes it difficult for them to have the patience to deal and talk to the rural customers (Journal of Digital Banking 2019). Through the power of AI, banks can come up with natural regional language processing-based AI-trained robots for training and talking to the rural customers in regional language (Paul 2019). These robots explain various banking products offered by the bank, the robots can also explain the amount of debt rural customers have and even offer suggestions on the need to save (Siddiqui and Siddiqui 2017). AI-trained robots can become financial advisors to rural households (Deloitte 2018b; Paul 2019). As a result, AI is helping a lot to allow previously vulnerable groups to be able to access formal financial services (Wang and He 2020).

Additionally, some customers can access banking services through their mobile phones, where they can transact even while at home in the remote parts of their countries as long as they are connected to mobile networks. Furthermore, the use of AI can help a lot in account opening as individuals can open accounts or deposit through the use of phones (Paul 2019; Wang and He 2020). The use of blockchain has also allowed usability of accounts to be more effective; it takes approximately 10 minutes to transfer money which is faster than the conventional means mainly used in developing nations (Paul 2019). When using blockchain technology in digital finance payments, there is no need for payments to go through the national payments system and as a result, there is no need for physical branches. This makes payments more feasible as the cost of the transfer is the percentage of the value of the transferred (Paul 2019). On some instances, AI can facilitate quantitative trading. AI-powered computers can have a deep analysis of large and complex data sets very fast and more efficiently than human beings. This will result in automated trading which saves valuable time (Wang and He 2020).

## 2.1.4. Fraud Detection and Cybersecurity

Ramping up cybersecurity and fraud detection efforts is becoming a necessity for any financial institution or bank because of huge quantities of digital transactions which are carried out via online accounts every day, sometimes through mobile phone and applications (Lopes and Pereira 2019b; Paul 2019). AI is playing a big role in the improvement of security of online finance. The ability of AI to offer this kind of security to online finance makes it possible for the people at the bottom of the pyramid concerning financial inclusion to be able to participate in the formal financial sector (Reim et al. 2020). Further, fintech companies are using AI applications to advance consumer protection and user experience, manage risk, detect fraud in many countries (Paul 2019). Various national stock exchanges in many countries are contemplating the use machine learning to identify market patterns to improve monitoring and prevent manipulation of its high-frequency trading (HFT) markets (Journal of Digital Banking 2019; Deloitte 2018b). In reality, AI-enabled cybersecurity systems are increasingly being used to guard against and prevent possible security breaches. In addition, AI is influencing wealth management through robot advisors that provide automated financial planning services like tax planning advice, insurance advice, health, investment advice and many other crucial services (Journal of Digital Banking 2019). The HDFC bank of India is using AI for its Mobile Banking App, and On Chat, which makes use of Natural Language Processing where users can interact, confirm and pay for services within chat (Paul 2019).
