Can AI Revolutionize Financial Consumer Protection?

Charina De Vera-Yap: BOB has been providing real time 24/7 access to the BSP redress mechanism since 2020. And using BOB as our main platform for complaints escalation allowed us to observe and respond quickly to emerging financial consumer risks. So that in itself provided a better customer experience when they interface with financial regulators such as the BSP.  

Lamis Daoud: Artificial Intelligence seems to be popping up in conversations everywhere. At home. At work. With friends. Chat-GPT is something of a household name now, and the limitless possibilities of ai and machine learning are profoundly exciting and terrifying all at once... Perhaps the reason why we’re all so interested in it to begin with.

Now, let's zoom into the financial sector. AI is reshaping not only the way financial services and providers operate, but also the way financial authorities operate, driving efficiencies and sparking innovation.  

This raises big questions: can AI be used to protect financial consumers? And if so, how and to what extent? Can it go a step further to help financial authorities improve consumer experiences and the value consumers derive from financial services?  

Hello and welcome to inclusive finance frontiers, a podcast by CGAP. I’m your host, Lamis Daoud.  

Back in 2019, the then-governor of the bank of England, mark carney, said that the bank received 65 billion pieces of data annually from companies that it oversaw… and that reviewing it all would be like each supervisor reading the complete works of Shakespeare twice a week… for every week of the year. And as enlightening as that sounds, it is overwhelmingly untenable. Carney went on to say that financial regulators would need to adopt ai techniques in order to keep up with the rising tides of data flowing into their systems.

Fast forward to today. The amount of data has ballooned, and the use of AI has risen along with it.  

Eric Duflos has been working on financial inclusion for 30 years and leads consumer protection work here at CGAP. He walks us through some of the benefits ai brings to financial providers and consumers.

Eric Duflos: AI has been around for quite a while. It's not completely new, but obviously, this is accelerating at light speed. If we look at providers of financial services, AI can play a very big role for them. For example, it can improve efficiencies in terms of delivering financial services, in terms, also, of onboarding new consumers. So through higher automations, providers can become more efficient. That means improved bottom lines, but that also means expanding financial inclusion to new consumers.

They can also use AI for predictive analytics that can help them sometimes warn consumers who are not in the best position. For example, it can predict whether some consumers are running some specific risks. So many roles that AI can play for providers, but it can also help consumers a lot. For example, consumers can get products and information that is more tailored to their needs through AI. It can also improve the way they can identify themselves, making it much easier for them to get formal financial services, for example, in terms of KYC.

Lamis Daoud: Today, most financial institutions acknowledge that they’re deploying ai to support areas like risk management, loan origination, income verification, fraud reduction, and compliance and auditing processes.  

On the consumer side of things, you might recognize ai chat boxes popping up in the bottom corner of all kinds of websites you visit, from shopping to banking. These chatbots not only help answer consumer questions, but they also help gather and analyze data which can provide more insight into consumer experiences and issues.

One of the earlier adopters of this kind of tech is the central bank of the Philippines, or B-S-P. It launched a chatbot – named bob, for “BSP online buddy” – in 2020 amidst the Covid-19 pandemic.  

We spoke with Charina De Vera-Yap, a managing director at the bank. She works in the financial inclusion and consumer empowerment subsector.

Charina De Vera-Yap: The chatbot journey is actually an offshoot of BSP’s financial inclusion journey. So as BSP’s financial inclusion initiatives are gaining ground, more and more Filipinos are of course becoming financially included in terms of account ownership. With that, we saw the need to also democratize access to a redress mechanism, especially with the emergence of digital financial products. We wanted to further enhance our group's operational efficiency in terms of handling complaints.

Lamis Daoud: If fintech – short for financial technology – enable and optimize financial services like online banking, digital payments, virtual currencies and more… then suptech – short for supervisory technology – is software that automates, manages and improves the supervisory operations of all that fintech.

It’s used by regulatory bodies that are trying to ensure that financial regulations are being met. Together with AI, suptech can be a game-changer for the industry.

Lamis Daoud: Suptech encompasses supervision, regulatory compliance and market analysis, and a lot of automated reporting. It thereby reduces a lot of reporting time, increases data granularity and looks for patterns and trends to help inform new regulation. This ends up improving the flow of information that supervisory bodies have, shortening the time it takes to process it, and thereby, gives financial supervisors and regulators the opportunity to take quick action to mitigate consumer harm.  

Juan Carlos Izaguirre is a senior financial sector specialist at CGAP. He leads the Responsible Digital Credit work and is an expert in the regulatory and supervisory space in digital finance ecosystems.

He explains how suptech helps streamline necessary supervisory processes, like electronic data reporting and analytics.

Juan Carlos Izaguirre: Qualitative data was not typically used by supervisors or analyzed by supervisors because it was too complicated to find different types of patterns in qualitative data that consumers would have available for supervisors to use, we can now use natural language processing to analyze all these type of unstructured type of data. And we are using these type of tools, natural language processing, to help go through large volumes of non-standardized qualitative information.  

Lamis Daoud: In essence, suptech helps supervisory authorities improve their effectiveness, communicate and gather insights directly from consumers and promote trust in the financial institutions and markets that they oversee, and that’s pretty powerful.

Charina De Vera-Yap: BOB has been providing real time 24/7 access to the BSP redress mechanism since 2020. And using BOB as our main platform for complaints escalation allowed us to observe and respond quickly to emerging financial consumer risks. With BOB, this greatly shortened our turnaround time when responding to complaints because prior to BOB, we were doing everything manually. So that in itself provided a better customer experience when they interface with financial regulators such as the BSP, and also with the complaints information and data captured through BOB, we regularly utilize that information to inform the legislative measures that the BSP proposes in the Philippine congress.  

Lamis Daoud: Powered by suptech, people can engage with bob via webchat on the central bank’s website, via text message – or SMS– and on messenger on BSP’s Facebook page. According to BSP, BOB has revolutionized the way the central bank receives and processes complaints to financial service providers, who are mandated to respond to complaints within a designated time frame. In 2022, just two years after launching the chatbot, 75% of the bank’s customer complaints were processed via bob.

That same year, the financial consumer protection act – a landmark piece of legislation passed by the Philippine congress – was signed into law.  

Charina De Vera-Yap: When we were discussing or that bill was being deliberated in congress, it benefited from the inputs that we provided with regard to complaints data. These complaints data came from the database of BOB. So that also helped us push and also expedited the passage of the law.

Lamis Daoud: Matt Grasser is co-head and chief technologist of the Cambridge Suptech lab which is housed at the Cambridge center for alternative finance at the university of Cambridge judge business school.

Their mission is to accelerate digital transformation of financial supervision. They do that through research, capacity building and education, application incubation, technical assistance, and ecosystem acceleration.

When the central bank of the Philippines wanted to build out their chatbot, they collaborated with an early version of the Cambridge Suptech lab.

We spoke with Matt to learn about how it all came to be.

Matt Grasser: We actually met the BSP colleagues back in 2017 before there was a conceptualization of whether a chatbot was a thing or whether there was a complaints collection as a priority or these sorts of things.

We were heavily focused in this conversation on what would be most impactful for things like financial inclusion or more equitable financial services in the Philippines. So we were highly aware of making sure that this was done in an inclusive manner. It wasn't just a system that's only available on smartphones. It wasn't something that's [00:07:00] only available for really digitally advanced colleagues, but also incorporated additional channels like SMS, Viber in the Philippines, voice calling things like this that worked for also people without as advanced technology.

Lamis Daoud: But it wasn’t just a technological access problem they needed to solve for. To make bob more inclusive, they needed to account for language as well.

Matt Grasser: We also saw the need for multilingual support, and not only multilingual, but mixed language support. So in the Philippines, many people speak Tagalog, many people speak English, and many people speak in their day-to-day language and in conversations over SMS or text-based systems, a mixture of the two, which is often referred to as Taglish. And so what we needed to do is make sure that this wasn't just an English-based chat chatbot solution, but was something that was adaptive and inclusive, and accessible for the complainants that needed to submit complaints to the central bank. So with all of these ingredients in mind, the value proposition for the authorities, the needs of the market, and the customer.  

Lamis Daoud: And how do you do all that? With the help of AI. Charina describes how it’s put to use.

Charina De Vera-Yap: AI allows the BOB chatbot to understand and appropriately respond to financial consumers in a live conversation. And with natural language processing, BOB is also able to converse with consumers in both English, Tagalog or Taglish. With this multilingual capacity, consumers can talk to BOB in the language that they are most comfortable with.

Lamis Daoud: But AI has already proved to be a double-edged sword. With all this positive, inclusive use, it is also important to consider the challenges, limitations, and potential threats that the use of ai can bring.  

Eric Duflos: I'm quite concerned about the scale-up of frauds. We conducted a research at CGAP a couple of years on the consumer risks from digital financial services, and we found that frauds were increasing really fast, as well as the misuse of data, and I'm pretty convinced that AI could accelerate this trend. Just one example that I'm very concerned about is what we call voice cloning. There has been examples of your voice being used by fraudsters to send wrong messages.  The other one that is probably more famous in terms of risks is the risk of discrimination. And we know that different algorithms can discriminate against specific groups of customers, based, for example, on their gender, ethnicity, race, or sexual orientation.

Lamis Daoud: This makes it even more important for financial authorities to be proactive in addressing such looming challenges and threats. While data privacy and security are always top of mind, it is important for financial supervisors to allocate resources to guide the development of an AI tool to avoid the very real risk of exclusionary practices.  

Juan Carlos Izaguirre: Sometimes, algorithms may incorporate or reinforce biases in discrimination that are brought to technology and continue based on the existing human type of biases. Also, sometimes, supervisors may rely too much on data analytics and not consider some quantitative aspects or qualitative aspects that the human touch bring with them. The risks on data and representativeness come because some data gathered through artificial intelligence may not represent all consumer segments.  

In terms of data privacy and security, some of the tools may heighten cyber security or data security vulnerabilities. There may also be some dependency risks, like tools that depend too much on the usage of cloud computing or the use of algorithms by third-party providers, and therefore, this dependency may lead to certain risks in terms of supervisors using a specific concentrated type of provider. So these are just risks that supervisors need to keep in mind when they use suptech solutions so that they address them in a good manner whenever they put them in place.

Lamis Daoud: At the end of the day, responsibility and intentionality are key drivers to ensuring the appropriate application of AI.

The growth of ai is happening at rapid speed. For financial authorities using the technology, it takes a proactive approach to expand on inclusion in digital finance and guard against misuse and unintended consequences.

My colleagues at CGAP explain.  

Eric Duflos: I think that on the one hand, technology can play an incredible role to help supervisors improve financial consumer protection, but at the same time, they need a whole range of tools that are not ai-driven, also to do this job, such as very good market monitoring tools, but also, a very sound regulatory framework.

Now, when we think at CGAP about financial consumer protection, we often refer to the notion of a responsible digital finance ecosystem, and we like to talk about ecosystem because one consideration for supervisors is that they need to work with other actors in the ecosystem if they want to fully utilize AI. And just to give a couple of examples, we often say that for a digital finance ecosystem to be more responsible, we need four components, which we call the forces, so we talk about customer centricity, we talk about collaboration, we talk about capability, and we talk about commitment.

Collaboration is absolutely necessary for supervisors to be successful in their use of AI and there are many different levels of collaboration that can be helpful, and one of them that I want to mention is the collaboration with other authorities working in the same country. Because what we're seeing is that, with the expansion of digital finance, with the expansion of the use of data, and the expansion of use of AI, which is also relying a lot on consumer's data, there is a need for more authorities at the table to make these financial services more responsible.  

Juan Carlos Izaguirre: Digital finance is evolving so, so fast that I think it's imperative for market conduct authorities and other stakeholders to take a really proactive, preventive, and forward-looking approach to consumer protection, and to understand, assess, and act on the risks that they observe in the market. Artificial Intelligence in suptech is precisely a tool that can help stakeholders to be at the forefront, to be able to take quick action and identify the issues that are happening almost at real time in the digital financial sector.  

I think it's very good to see examples such as BOB. Making AI tools tailored to the specific context of different jurisdictions are key, and making the efforts to make them appropriately tailored to these specific jurisdictions is an effort that makes all sense and it's important to prioritize. Also important is to take into account how AI may be addressing some particularities of customer segments that are affected, and in some cases, marginalized, or even persecuted. It's very important to make sure that data that is sensitive, that is quite personal, is adequately addressed and protected, and that issues of vulnerable segments are addressed whenever AI and suptech power tools are being deployed in the market.  

Lamis Daoud: So what’s next for bob? More automation for more information… leading to more inclusion.

Charina De vera-yap: We are looking forward to having a more user-centered design, more intelligent and agile exchanges within the BOB platform and more informative and timely analytics. We are also envisioning to automate more bob-related processes and integrate other data from other sources such as social media so we can have a better view of consumer issues. We're trying to look at the possibility of BOB having an interface or which can be integrated in other BSP-wide systems so that we'll have a better grasp of the data that's out there being collected also by the BSP through other systems.  

Lamis Daoud: And what’s next for the future of suptech? Matt sees the technology having a domino effect. That can lead to problems being solved that we might not have even thought of yet.  

Matt Grasser: In our experience, what we've seen is that one suptech solution begets another suptech solution that when you create something to, for instance, streamline the ingestion or collection of data, whether that's complaints or whether that's other forms of data, once you have that information internally in your systems and it's cleaned, processed, stored properly, and available to the teams of supervisors that are using this type of data, that system resolves one major, major pain point and opens a whole suite of opportunities for additional over-the-top composable solutions.  

Lamis Daoud: It’s a future we’ll be monitoring closely.

That’s the end of our podcast episode today. I’d like to thank our guests, Charina, Matt, Eric and Juan Carlos for taking the time to speak with us. And thank you so much for joining us in this episode of inclusive finance frontiers. I’m your host, Lamis Daoud.  

For more information on our work in financial consumer protection and inclusive finance, please visit our website at cgap.org. And if you liked this episode, please share it with friends and colleagues so more people can hear how inclusive finance is transforming lives.