I asked ChatGPT about how AI could help accelerate financial inclusion
Image credits: Accion and Australian Research Data Commons

I asked ChatGPT about how AI could help accelerate financial inclusion

In my mid-teens, I wanted to create a text-based game along the lines of the then popular video game, “Prince of Persia”. Programming languages I knew – Borland C, for example – could not achieve the dynamic interaction needed. My investigations led to LISP, a language that allowed the manipulation of lists of text in unique ways based on interaction with the user. So, I taught myself LISP and “fuzzy logic” concepts and built the game. I found, by sheer chance, that this was my first attempt at building “Artificial Intelligence”.

It worked, somewhat – and the experience with LISP opened an endless list of possibilities. I used LISP for computer graphics applications and even dynamically creating a database of patterns of mineral compositions based on outputs from a mass spectrometer which was available at IIT Delhi’s IDDC lab that I could use as part of my high school.

Yes, I didn’t get out much in the hot and humid New Delhi summers!


So, what does this have to do with financial inclusion and digitally enabled economic development?

Before I begin, let’s define #artificialintelligence (AI). Back in the 1950s, John McCarthy coined the term. His 2004 paper defines AI most eloquently:?“It is the science and engineering of making intelligent machines, especially intelligent computer programs. It is related to the similar task of using computers to understand human intelligence, but AI does not have to confine itself to methods that are biologically observable.

According to the?World Economic Forum, AI can democratize financial services with the appropriate infrastructure, data-sharing environment and ethical framework. Here is another great paper that talks about the role of AI in attaining the sustainable development goals. AI can be used for the broader development agenda, for example Raesetje Sefala?has been building algorithms that flag poverty hot spots and developing data sets she hopes will direct aid, housing or clinics to the areas that need it most.

Coming back to #financialinclusion and impacting the lives of the underserved smallholder farmers and micro businesses at scale, I can think of three types of entities that leverage AI tools.


1.????For financial services providers:

  • A key challenge to be addressed when rolling out solutions, especially for underserved people, is getting over the literacy barrier – how do we get people to interact with text on a mobile device? Conversational AI, such as chatbots and voice assistants, have the potential to facilitate communication and access to information for people with lower literacy levels by providing an alternative to text-based interfaces. By using #naturallanguageprocessing (NLP) techniques, these technologies can enable people to manage their loans, insurance, and savings accounts or troubleshoot problems through spoken or written language, rather than requiring them to read and understand text. Imagine an intelligent interactive voice response solution offered via a WhatsApp call to the financial services provider’s conversational AI platform.
  • AI tools can be used to analyze “big” and “small” data from various sources dynamically to assess the creditworthiness of potential borrowers and address the needs of even thin-file clients. It is important to avoid potential for algorithmic bias.
  • AI tools can help identify patterns that could indicate fraudulent activity and flag potential security risks. This is already being done in many large financial institutions.
  • Dynamically recommend optimization of digital channels for customer and staff engagement through rapid analysis of usage patterns. This can help organizations make investment decisions regarding marketing to promote usage for example.
  • Identify patterns and trends that can inform the design and implementation of financial services, such as by identifying the most effective interventions for addressing specific challenges or by identifying areas for improvement within internal processes.
  • AI can be used to analyze data about customers' financial needs and preferences to provide personalized recommendations for financial products and services.


2.????To increase efficiency of #supplychains and #digitalmarketplaces:

  • AI tools can be used to analyze patterns within supply chains from across various sources, via secure APIs, to dynamically manage the velocity of movement of goods and predict where needs may arise for goods distribution planning.
  • Offer micro businesses personalized inventory stocking recommendations and just in time financial products based on their cashflow patterns, coupled with wider economic and environmental data.
  • Use AI to match e-commerce marketplaces with small businesses to help them buy inputs and sell outputs.


3.????For #impactinvestors :

  • AI can help analyze data from a variety of sources to better understand the social and environmental impacts of investments within the broader country context of each of the portfolio companies, which can help investors to identify opportunities to better understand the impact of their investments over time.
  • Identification of patterns and trends to inform the design and implementation of impact investing strategies, such as by identifying the most effective interventions for addressing specific challenges or by identifying areas for improvement.
  • AI can be used to analyze data about investor preferences and goals based on their wider investment strategies to provide personalized recommendations for impact investments that are aligned with their values and objectives.
  • AI-powered tools can be used to educate and engage investors about the benefits of impact investing and to provide resources and support to help them get started.


So, what would be needed to make this work?

  • Management sponsorship for experimentation and innovation and a technology team that can help realize the potential for AI.
  • An AI solution from an existing AI platform like OpenAI, or IBM’s Watson, etc. The good thing about GPT-3 models (and beyond) is that they are designed to learn from the data fed to them. They can then generate new data like what has already been received. You’ll find more detailed information on AI models?here. Also, a “pre-trained” model doesn’t need to learn everything from scratch. It can use old knowledge and apply it to new tasks.
  • Data, Data, and Data! Without adequate datasets that can be used by the AI to learn from, it is difficult to create an AI tool that can offer responses which are accurate with a high level of probability. ?The power OpenAI’s latest (Nov 2022) GPT-3 model, davinci-003, is that it reduces the amount of human supervision to fine-tune the model - thereby reducing the time to launch new services based on AI.
  • A technology platform that can compute at a large scale and velocity. These are available from cloud providers including Microsoft Azure, Infosys Nia, Google Cloud AI, and Amazon Web Services, amongst others.


In the end, this is a #digitaltransformation project for most institutions, which is not a trivial exercise. However, those who use AI based solutions and frameworks from the get-go may have a stronger chance of attaining business goals quicker than planned and leapfrog competitors.


Separately, it is important to note that AI technologies have been around since the 1950s and are only just starting to become technologically viable and commercially feasible. AI, in its current state, is best thought of as a solution that supports people rather than replace them since we are in the early stages of this technological revolution.


We live in exciting times!


Footnote: many parts of this blog were generated by ChatGPT and edited by me.

Gregory Ubigen

Payments, Pricing, & Product Strategic Leader / Built several fintech businesses & propositions from ground up.

2 年

Thanks for sharing, Prateek. Were you wearing a turtle-neck and working in a garage, while building the app in your pre-teens? ?? Agree: there are a lot of opportunities for #AI to help scale #financialinclusion To your points about supply chains and digital market places: I think there is an opportunity to bring together device manufactures and financial institutions (for example OPPO Indonesia and PT Bank Mandiri (Persero) Tbk. in Indonesia) to offer subscription based devices that target SMEs with built-in inventory management and ecommerce solutions that are tied to the financial inclusion offerings of the bank, but that leverage the power of AI to achieve some of what you have mentioned above. One can bring in potential ecosystem players that supply goods to these SMEs to help drive adoption. And it could be a very exciting proof-of-concept to look at, wouldn't you say?

Raúl Gómez-Velásquez S. ? ???? ????

Full-time Husband and Father | Board and C-Suite advisor | Strategy | Digital Innovation with a Gender Focus | Financial Inclusion | Fintech | People, Processes, IT and + | Inspiring Speaker on Inclusive Finance ??

2 年

Excellent article Prateek! I am only concerned that these disruptive technologies will eliminate the relational model that has been the essential feature of microfinance and could fuel a frenzy for over-indebtedness of the most vulnerable populations.?In developing countries, 29% of adults still do not have access to an account, according to Global Findex 2021, where the paradox of over-indebtedness in urban areas (excess supply of microcredit in the most accessible areas with better level of connectivity) coexists with a population in rural areas with little access to financial products and services.?RGV

Simon Aderinlola

Fintech, Inclusion & Startups specialist. Ex-Accion, Telco & Bank GM | Co-bootstrapped an Agent Network to Interswitch acquisition | Regulator engagement, Policy & Digital Fraud mastery.

2 年

Rethinking the footnote, Prateek. Lovely article with your flavour, with great prospects for the co-author stated :)

要查看或添加评论,请登录

Prateek Shrivastava的更多文章

社区洞察

其他会员也浏览了