Going digital and harnessing data are now widely acknowledged as strategic differentiators for the financial services industry. Many financial services institutions are turning to machine learning and artificial intelligence to accelerate and augment their business models.
In conversations with bank executives, a common concern we hear is: How do I capture real world benefits of AI under strict regulatory scrutiny?
In Q1, we saw a number of trends emerge amongst bank and non-bank lenders. The Monetary Authority of Singapore (MAS) announced the development of a fairness assessment methodology in credit risk scoring and customer marketing under the Veritas Initiative. This is a significant step forward in shifting the governance conversation from principles to actionable steps. We see this as part of a broader push towards implementing responsible AI, and an important area in which to stay ahead of regulatory developments.
The New York State Department of Financial Services also released its findings on its investigation into the Apple Card program, which concluded that there was no discrimination based on sex. However, there were deficiencies in customer service, such as the degree of transparency with credit terms. There was a common misconception that spouses are entitled to equal credit terms from credit card issuers if they “shared finances”. This shines the spotlight on the importance of explainability and transparency on perceived fairness. A customer’s experience of fairness can be just as important as the final outcome.
Another question we get from executives in the financial services industry is: What are the focal areas in banking for AI solutions? Or more candidly, where do we get the real bang for the buck?
The use of natural language processing (NLP) is one area that is gaining considerable momentum for use cases such as process automation, document processing and intelligent news screening. We are also seeing both bank and nonbank players harnessing alternative data for applications such as credit risk scoring and hyper-personalisation.
We have just published a report highlighting the key trends we observed within FSI during the first quarter of 2021. Here we are share an extract from the report:
Trend #1: Increased demand for natural language processing within financial services.
Natural Language Processing (NLP) has the potential to free up significant middle and back office resources for the financial services industry (FSI) to fundamentally redesign its structure and provision of services in light of disruptive innovations.
NLP enables machines to process human language in the form of text or voice data, and decipher its full meaning, complete with the speaker or writer’s intent and sentiment. By combining computational linguistics—rule-based modeling of human language—with statistical, machine learning, and deep learning models, NLP has the power to drive applications that can translate text from one language to another, respond to spoken commands, and summarize large volumes of text rapidly—all in real time.
NLP is a highly in demand area that banks are approaching AI solution providers for. In Q1, we saw FSIs use NLP for:
- Intelligent news screening: Assigning ‘relevance scores’ to news articles in order to rank them and focus a team’s attention on the most pertinent ones. For example, tracking content on an emerging sustainability topic, such as ESG, or articles with information that might lead to a potential business deal.
- Report examiner: Extracting information from PDF files, such as annual reports. Information can be extracted from tables and unstructured paragraphs, to help augment analysts in their proposal generation or due diligence processes.
- Document processing: Identifying fields from contracts or forms across multiple different formats. AI engines can ‘understand’ the contents of documents and accurately extract information and insights, as well as categorise and organise documents.
Trend #2: New nonbank, cloud native players are harnessing alternative data for credit risk and hyper-personalisation.
Southeast Asia is home to an estimated population of 290M unbanked, or underbanked individuals. Coupled with the crippling socioeconomic effects of the pandemic, progress towards financial inclusion goals has taken a backstep.
It is therefore unsurprising that we are seeing new non-bank players in the region emerging to bring credit to those without credit scores through digital offerings such as e-wallets, credit lending, Peer-to-Peer (P2P) lending and equated monthly instalment (EMI) financing. These cloud-first enterprises are harnessing alternative data sets, such as telco data, in decision making for credit risk scoring and loan decisioning models. They are also tapping on hyper-personalisation to embed financial education into offerings, such as the use of gamification to help individuals better manage their finances. By relying on the fusing of data sources, these organisations are able to direct hyper-targeted, contextual, personalised recommendations to users.
Trend #3: Fairness and explainability becomes top priority as regulatory scrutiny is anticipated.
As the momentum for AI adoption increases, there is a greater awareness of the risks associated with deploying AI systems that violate legal, ethical, or cultural norms.
Beyond the broad articulation of AI governance principles and frameworks, emerging signs are pointing towards greater AI regulation to manage these downsides. The EU has its General Data Protection Regulation that will impact how AI technologies use data. Governing authorities such as the UK Information Commissioner’s Office has taken a step further to publish a draft guidance on AI Auditing Framework. The Monetary Authority of Singapore (MAS) has developed a Veritas initiative to develop a fairness assessment methodology in credit risk scoring and customer marketing.
Whether it is governance through soft law, policy guidance or audit frameworks, organisations need to be ready to demonstrate the ability to uphold trust and confidence in their use of AI. As such, fairness and explainability is top of mind for many data science leaders in the sector who are looking for new ways to operationalise the prevention of unintended bias and ensure model explainability.
Contact us to learn more about AI solutions for the financial services industry.