Artificial Intelligence and Financial Inclusion in Myanmar
Updated: Jul 16, 2018
Recent advances in technology and the availability of computing power has made artificial intelligence a reality. With the help of AI, financial organizations can now deliver better customer experiences, streamline their compliance processes and reach new customers on a massive scale. So, what can AI do for emerging countries like Myanmar? More specifically, can AI be used to improve financial inclusion in Myanmar?
Myanmar Country Brief by UNCDF
In this country brief of Myanmar from the UNCDF, the mobile subscription rate is 93% in 2017, compared to 6% in 2012. In the same time period, the financial inclusion rate has remained almost the same (23% in 2013 vs. 26% in 2017). With such a high mobile subscription rate, embracing AI and digital financial services may seem like the only solution to improving financial inclusion.
What is AI?
Artificial intelligence is defined as the capability of a machine to mimic human behavior, from playing chess to self-driving cars. How can AI be utilized in the finance industry? Firstly, it is important to remember that in this case, AI should be used to augment human behavior, not replace. AI's ability to handle large datasets is essential, especially in such a data-driven industry. AI also provides great return on investment in 3 key areas:
1. Cost Savings
Reducing the costs of delinquency by identifying bad loans at acquisition. Developing alternative credit scoring platforms that make use of data such as payment history for basic services to assess an individual’s payment capacity and credit-worthiness. Such models allow for unbanked individuals to get better access to mainstream financial services.
2. Revenue: Reaching new markets and customer segmentation
Identify micro-segmented borrowers and design long-term loan products that take into account their finances or business performance to help them climb the credit ladder.
3. Differentiation: Automation, personalization, seamless experience
Chatbots that can automatically address customer’s queries and complaints
Reduce fraud by using facial recognition to authenticate customer’s identity during transactions
Track user activity to tailor financial offers based on individual needs
In the African financial market, AI has been used extensively for credit assessment. Branch and Tala, two lenders in Kenya, have been using credit assessments and behavioral data from mobile phones to provide mobile credit to individuals. Information such as demographics, social media information, call logs, phone make and model are used to form a profile of the customer. South African lender Lulalend has been using machine learning and predictive analysis to assess business health and predict future income of their borrowers. Combined with an online accounting platform, they can determine creditworthiness. Because their system is automated, Lulalend decision-making process is much faster than other lenders and have one of the lowest default rates in the market.
Tala loan app
Lulalend online lending for SMEs
Abe AI, a US-based banking solution provider, and South African bank Absa, have partnered to create a service that can predict customer savings activity and spending. Using data collected on financial habits and behaviors, they can predict cash flow, future purchases and provide overdraft protection, helping customers practice healthy financial behavior.
According to a recent Financial Times survey of 30 leading banks, many in the finance industry are looking to turn to AI to help reduce cost through the use of chatbots for problem-solving, inquiries and customer service. South Africa-based FinChatBot has been helping insurance companies lower cost by using chatbots for basic inquiries. US-based Juntos is partnering with financial service providers in an effort to change customer behavior by providing them with individual goals – whether it’s to increate their savings balance or reach a business goal – that are based on analysis of SMS conversations.
Voice-recognition and text-to-speech applications can help reach potential customers who may be illiterate or having hearing or speech impairment.
28% of Google searches in India are conducted by voice. These applications can help simplify terms and processes that may have been difficult for them to understand or navigate before.
Using AI to respond to basic customer inquiries creates a more efficient and seamless customer experience. It is likely that the majority of the customer inquiries will all be the same questions - how do I apply for a loan, what are the available loan terms etc. Normally, customers would have had to call the financial service provider on the phone during office hours for any information. Through chatbots, customers can ask these questions any time they want and as many times as they want.
The data collected through AI could also help providers identify potential markets by combining transactional data and contextual geographical data such as population density.
Natural language processing may be an issue as AI may have problems processing the local language.
Having the data itself is not enough. There needs to be a proper system for collecting, housing and organizing the data. It is also important to make sure that the relevant parameters and predictors are collected as well.
Data privacy may also prove to be a challenge. Customers are entitled full knowledge of their data rights. Any information that they request on how their data is stored or used must be provided to them in simple and understandable language. Having the proper infrastructure, as well as regulations, is needed to ensure this.
There is also a fear of artificial intelligence by some who feel that AI is a risk to data privacy or could potentially lead to the loss of jobs. Most of this stems from a lack of understanding or trust of technology. Transparency and positive user experiences can help build a relationship with users, which is crucial for AI to work.
The lack of capacity and talent can be a huge obstacle for organizations, especially in Myanmar. Many providers simply do not have IT or Business Intelligence capacity required to begin using AI. Using AI may also require changes in the way the organization works, from processes to policies. There can often be some resistance to change, either from employees or management.
"Pite Pite" Financial Chatbot in Myanmar
ThitsaWorks has recently launched a financial chatbot in Myanmar named Pite Pite (a slang for money). Our primary goal is to provide financial education and financial access. Pite Pite’s functions include:
Repayment and Savings calculator
Service eligibility check
Financial institution locator
Product suggestion based on user profile and preferences
A marketplace that can match users with available products based on a questionnaire
A pre-application that can be filled out and sent to financial institutions
Pite Pite was built using the Microsoft AI platform. Through this platform, we were able to utilize tools and services such as Azure Machine Learning, Visual Studio Code Tools for AI and Bot Framework.
Why AI now?
Computer processing power and data storage have become considerably more powerful and affordable. Cloud infrastructure has reduced costs to a fraction of what it used to be. Open-source algorithms have also been crucial for start-ups.
AI processes data better and faster than traditional systems. AI can process data in real time at a larger scale and provide more accurate predictions.
The data is already available but will need to be cleaned, organized and stored properly.
Financial institutions may face competition from neobanks and superplatforms.
There is a noticeable demand for tailored, real-time servicing and feedback.
Here is a story of a user of Pite Pite financial chatbot to access financial information. Users like her are the reasons behind why we do what we do.