There’s much anticipation that combining new data and artificial intelligence will transform the pace of financial inclusion and the scale at which it’s possible.
It’s easy to see why. AI is being used to rapidly perform tasks that would otherwise require endless hours of human labor. It can, for example, sift through and analyze enormous volumes of non-traditional sources of data to expand access to credit. It can digitally verify identities, detect fraud, and improve customer engagement.
Institutions view AI as a tool that could greatly expand the number of customers they serve without having to scale their workforces by anywhere near as much.
This reasoning is compelling, but it obscures a wider point: the technologies that drive automation require people with specialized technical skills to implement, monitor, and refine them. These people are, for now at least, in very short supply.
We’re the authors of an upcoming report investigating the risks that will arise as financial institutions increasingly embrace AI (to be published by the Centre for the Study of Financial Innovation [CSFI] later this year). Our research suggests that one of the most fundamental of these risks – and a threat to institutions, their customers, and the financial system as a whole – is a pervasive skills gap.
The Data Scientist Problem
Last year, the CSFI’s “Finance for All” report surveyed respondents from the financial inclusion sector on the 20 greatest “banana skins” – risks financial inclusion professionals believe pose the greatest challenges for service providers. At the top of the rankings was technology: the risk that the industry fails to take advantage of new technologies or implements them poorly.
Meanwhile, one of the fastest rising risks was talent: the ability (or not) of service providers to attract and retain suitably qualified staff. In 2016, it ranked 15th; last year it was 8th. But this ranking, if anything, understates its importance. In many ways talent risk lies at the heart of technology risk.
A report published last year by the CFI and the Institute of International Finance (IIF) found:
“Banks face a tremendous shortage of experienced data scientists and engineers, as well as AI experts, all of whom are needed to help convert big data “noise” into meaningful insights. Even relatively basic data skills are lacking in many major institutions, resulting in wasted resources, project delays, and squandered opportunities.”
Who are these people in short supply, and why do their skills matter?
Data scientists write the algorithms that underpin AI applications and determine which data should be used as inputs. They work with data engineers, architects, and operations specialists to test, implement, and monitor these algorithms on an ongoing basis for flaws and failings. Before doing any of this, they must work closely with the business in order to gain a detailed understanding of customer requirements. In other words, they define what it is the AI should do.
There is a global shortage of data scientists. A recent study by LinkedIn, for example, found that “demand for data scientists is off the charts” in the US. It identified a deficit of more than 150,000 professionals. The Chinese shortfall is predicted to reach 1.5 million vacant positions by 2020.
This shortage is especially pronounced in the developing countries, which tend to have higher levels of financial exclusion.
This shortage is especially pronounced in the developing countries, which tend to have higher levels of financial exclusion. For example, in 2017 it was estimated that there were just 300 to 400 data scientists in the whole of Thailand – and a current deficit of around 1 million people in the Asia Pacific region as a whole. A further concern is that the gap between leading AI centers and the rest of the world is likely to diverge rather than converge over the coming years. A report by the consultancy McKinsey found that increasing adoption of AI threatens to: “widen gaps between countries, companies, and workers.”
Traditional Financial Institutions Could Struggle to Attract Technical Talent
Data scientists and other AI specialists with in-demand skills are likely to have competing opportunities for well-paid and exciting work from many different industries. Nimble fintech start-ups with business models that use AI to rapidly scale financial services in previously unreachable markets may be an attractive destination. However, staff of Accion’s fintech investment funds report that talent acquisition ranks very high on the challenges their investee companies face.
More traditional financial institutions – from banks to MFIs – could find it even harder to sell themselves. A large number of respondents to the Finance for All survey were concerned that traditional institutions are not offering enough to attract and retain ambitious talent – in terms of compensation, challenging work, and advancement opportunities. A respondent at an MFI in Zimbabwe predicted, “There will be increased mobility of skilled staff to more challenging and rewarding institutions, as well as more technologically mobile institutions.” In particular, skilled technical people who might otherwise be interested in the industry are frustrated by legacy IT systems and infrastructure, which make it difficult to perform even basic IT tasks – let alone collect and analyze new kinds of data and collaborate with external AI providers.
Consequences of Not Attracting the Right People
For those service providers seeking to develop their own AI capabilities – particularly the largest financial institutions – the challenge is twofold. They need to attract a sufficiently deep pool of technical talent to avoid rushing solutions to markets that are buggy and vulnerable to outages and hacking. They also need to attract the right people – data scientists who understand the financial inclusion sector and can help build applications designed to meet the needs of customers, rather than building clever solutions in search of a problem. The need to find skilled technical people with relevant domain expertise makes a small pool of prospective candidates even smaller.
But it isn’t just the innovators that need to get this right. The majority of institutions that don’t intend to develop their own AI capabilities are also threatened by the absence of specialist knowledge – because they will increasingly form partnerships with fintechs.
To form successful partnerships, these institutions need to know which AI applications and third-party providers can be adapted to their business models – and, equally importantly, which should be avoided. They need to understand how to shape a strategy around AI: to what extent and in what ways they should rely on third party solutions, and the capital outlay and time needed to implement them. There are myriad other risks brought about by the decision to employ AI, from how to protect sensitive data to how to comply with regulatory requirements. It is difficult to mitigate these risks without specialist knowledge.
This is especially important in a climate where institutional survival may depend on the ability to adapt to new technologies. Alongside the fintechs that provide effective and useful solutions will be others that are prematurely hyped. Todd Farrington, Director at Symbiotics S.A. in Mexico, observed in Finance for All, “The proliferation of fintech solutions is promising, but many algorithms remain insufficiently back-tested, many companies young and overly ambitious.” Decision makers at partnering institutions who are under pressure need people who can demystify the rapidly changing technological environment.
There are also potential implications for market structures. The concentration of AI talent in developed economies could lead to more multinational financial institutions – or fintech start-ups that originate in developed countries – capturing large underserved markets and undermining the competitiveness of local operators. If foreign entrants gain decisive market share via first mover advantage, they’re likely to benefit from economies of scale that domestic competitors will find difficult to overcome.
If foreign entrants gain decisive market share via first mover advantage, they’re likely to benefit from economies of scale that domestic competitors will find difficult to overcome.
Prioritize Data Science Training
These are not questions with straightforward answers. The training of many more data scientists and other AI specialists needs to be prioritized, in financial institutions’ boardrooms and at a national level. The South African government, for example, said last year that it would train one million young people in data science skills over the next decade. Firms will have to get better at attracting technical staff, by improving incentives and overhauling antiquated IT infrastructure.
The crucial point is that the industry needs to prioritize bridging this skills gap immediately, and for the longer term. Otherwise, we may well find that the key factor limiting the benefits of the AI revolution in financial inclusion is not the technology or the data, but the ability to attract the right people.