Innovating responsibly in banking
Customer-centricity and proactive risk management are key to the consistent delivery of positive artificial intelligence outcomes
SINCE its launch in November 2022, ChatGPT – OpenAI’s generative artificial intelligence (AI) chatbot – has generated as much excitement as it has created alarm. It introduced the concept of generative AI to the public, which is broadly defined as a type of AI that uses massive datasets to stitch together new visuals, texts and audio content unique to a user’s request.
Development of this technology has been exponential since, with many new players jumping on the innovation bandwagon. For example, OpenAI rolled out voice and image capabilities in September this year, enabling users to engage in conversations and visual sharing with ChatGPT. Within the same week, Meta announced a partnership with Ray-Ban to develop camera and audio-equipped smart glasses connected to a virtual AI assistant.
Although AI has been around since the 1950s, significant improvements in computing performance and storage in recent decades have catalysed its application and adoption. As demand for the technology surged, it spurred the emergence of hyperscale data centres in cloud infrastructure services, as companies ramped up resources to enable AI workloads.
Synergy Research Group notes that corporate spending on cloud computing has ballooned, growing by an average of 42 per cent a year in the last 10 years to reach US$227 billion in 2022. This trend looks set to continue into the foreseeable future, especially as companies increasingly turn to the cloud to harness the power of generative AI.
Besides an enabling cloud-data architecture, a strong data foundation and robust governance processes are critical for companies looking to harness the benefits of AI, to ensure appropriate levels of data quality and confidentiality.
Particularly in the finance industry, appropriate adaptation of technological innovation is both a strategy and necessity. From automation of tellers to online and mobile banking, banks have sought to make customer experience more seamless and convenient.
BT in your inbox
Start and end each day with the latest news stories and analyses delivered straight to your inbox.
In 2021, UOB announced a S$500 million investment to ramp up its digital capabilities and accelerate AI-driven digital engagements across Asean. Over the years, the bank has scaled up the use of AI, incorporating it into many areas of its operations.
AI has enabled the bank to perform micro-decisioning and hyper-personalisation at an unprecedented speed, granularity and volume. It can deliver hyper-personalised insights, nudges and recommendations to its customers via its all-in-one banking app, UOB TMRW, by leveraging AI to analyse data based on these customers’ digital footprint, such as their personal spending and saving habits.
In 2022, UOB delivered over 110 million personalised insights to more than 2.5 million customers in the region. It also uses AI to scan through millions of daily transactions in real-time to identify cases of potential fraud and scams, and to alert customers to take actions to avoid financial losses.
Managing the dark side of AI
While AI can enhance customer experience, failure to deploy AI safely and responsibly can lead to adverse outcomes for banks and their customers, including non-inclusiveness, inaccurate decision-making and loss of sensitive data.
The emergence of generative AI has further heightened concerns over ethical and safety implications, such as intellectual rights and cybersecurity risks, leading to increased regulatory attention worldwide. Organisations are constantly being challenged to strike a fine balance between the use of AI to drive growth and improve customer experience, and managing the associated downside risks.
UOB is committed to deploying AI responsibly. Through the Fairness, Ethics, Accountability, and Transparency (Feat) Committee and Veritas Consortium, the bank has been one of the lead contributing banks, participating actively in the development of principles and an assessment methodology that guides the responsible use of AI among financial institutions.
Project MindForge is another example where UOB partners the Monetary Authority of Singapore and other industry and technology stakeholders to examine the risks and opportunities of generative AI for the financial sector. While some risks may be addressed by existing guidance and mechanisms such as Feat and Veritas, a new palette of risks is emerging that may require additional monitoring and responses.
UOB is also a proud signatory of the AI Movement Pledge to support the AI Ethics Movement, which is driven by the Singapore Computer Society (SCS) to shape a trustworthy AI landscape. The bank is honoured to be recognised by SCS as being among companies with the largest number of employees certified in the AI ethics and governance space.
One critical and sensitive area of AI application is decision-making, especially those affecting customers. To ensure that the output is robust and explainable, staff involved in the process must be equipped with the right skills, knowledge and experience to understand the limitations of the technology and to validate the output.
It is also important to remember that as technology and its applications evolve, so do their risks. While organisations should enjoy the benefits from AI applications, they must also stay abreast of its developments so that risks can be understood and managed.
UOB adopts a holistic approach to govern the use of AI, and has embedded guardrails in its business processes to proactively manage the associated risks. The bank has multi-disciplinary teams set up to identify, assess, mitigate and monitor risks across the entire course of systems’ design, development and deployment of solutions and processes.
This ensures a coordinated response from all aspects of the bank – its people, processes and technology. In addition to emphasising individual accountability, independent checks and balances are also established, and all staff underscore the tone from top management.
UOB’s approach to generative AI is to experiment with progressive and incremental rollouts that provide quick wins and important learning points, while limiting its exposure to unforeseen risks, particularly with regard to security and protection of customer data.
The bank is proud to be the first Singapore bank to deploy Microsoft 365 Co-pilot, a generative AI-powered office productivity tool which enables its staff to automate low-impact tasks and boost productivity and collaboration securely. Through Co-pilot, the bank monitors and manages risks that may arise and learns from the experience before expanding its application across the bank.
Amid the excitement surrounding Generative AI, it is important not to lose sight of classical AI, which will remain core to value creation. It is particularly suited to provide reliable and efficient solutions for many domains, such as product recommendations, risk management, optimisation, and decision-making.
McKinsey estimates that Generative AI can potentially unlock between US$2.6 trillion and US$4.4 trillion in annual economic benefits when applied across industries. Yet, this is still only 15 per cent to 40 per cent of the US$11 trillion to US$17.7 trillion of economic value that classical AI and analytics can deliver annually.
As stewards of its customers’ assets and information, the bank sees the importance of building trust and remaining vigilant to emerging risks while pursuing growth and opportunities to improve customer experience. Collaboration with regulators, industry partners and public stakeholders will pave the way for AI technology to create long-term, sustainable value for the bank’s customers and society, as the future of Asean continues to be built.
Richard Lowe is UOB’s group chief data officer. Alvin Eng is the bank’s head of Enterprise AI.
Copyright SPH Media. All rights reserved.
