Trust is one of the principal forces that binds society together; trust is also fundamental to banking. With artificial intelligence (AI) technologies set to reshape banking, financial institutions are raising critical questions about the degree of trust that should be conferred upon intelligent systems devoid of robust checks and balances. As AI evolves to become more sophisticated, and decision-making is increasingly driven by AI models, a key question I often ask myself in my role at a leading international cross-border bank is: how do we build and deepen trust in a technology whose decision-making processes are often opaque?
Banking on AI
Statista research projects that the banking sector’s spending on generative AI will surge to USD85 billion by 2030, underscoring AI’s expanding role in reshaping banking as institutions leverage AI technologies to enhance client experience, optimise operations, and drive innovation. In 2023, the banking sector led AI investments, trailing only behind retail in overall spending. The AI race seems to be in full swing.
But should AI adoption be a race at all? Neither the fear of trying, nor the fear of missing out should dictate our approach to AI. Instead, a thoughtful balance is essential – ensuring that AI deployment serves a human-centered purpose, adds tangible value, and benefits consumers and our wider society.
To say AI offers substantial benefits is an understatement. According to Accenture, banks can boost productivity by as much as 30% using generative AI over the next three years. At Standard Chartered, we have seen significant gains from AI deployment, leveraging it as a support tool, rather than a replacement for the creativity, problem-solving, and decision-making that define us. By automating routine tasks, AI has freed us to focus on what we do best.
For instance, we deploy AI to improve how we serve our Corporate & Investment Banking clients through client and frontline analytics that give insights for better working capital decisions, FX hedging, more efficient liquidity deployment and cross-selling recommendations.
AI also plays a critical role in risk management. Predictive analytics allows us to assess risks that could affect our clients from financial crime, ensuring we can take proactive measures to protect both clients and the broader financial ecosystem. In our Wealth & Retail Banking business, we leverage strategic partnerships utilising AI in credit lending decisions and enabling small businesses to access financing.
Bias as a Governance Issue
While AI has the potential to enhance inclusion, it is also known to produce biased results. Consumers are aware of these issues; 63% of consumers are concerned about potential bias and discrimination in AI algorithms and decision-making, according to the 2024 Zendesk survey.
To address these concerns, AI’s potential must be matched by robust governance. Bias arises when AI models are trained on historical data that reflect inequalities or when sensitive attributes, such as race, gender, or religion, are improperly considered in decision-making processes.
The use of sensitive data in AI models must be carefully controlled through strong governance to mitigate bias. This is why we have established the Responsible AI Council to guide our approach to both purchased and developed AI technologies at the Bank.
Governance provides a structured approach to identify, assess, and address biases throughout the AI lifecycle. To avoid unjust bias and other unintended consequences of AI, some key principles should guide its adoption:
- Accountability is critical. When AI systems make decisions, there must be an accountable individual overseeing the rules and processes governing those decisions. For us, this responsibility falls to process owners, ensuring that human oversight remains integral to our AI operations
- Explainability is another key aspect of responsible AI. Human oversight allows us to analyse, explain, and justify AI-driven decisions, which is essential for both compliance and client trust.
- Transparency is crucial to earn and retain trust. Clients should be informed when AI is involved in decisions affecting them, such as loan approvals or insurance premiums. Clear communication about AI’s role helps clients understand and trust the technology.
- Data quality is paramount. AI is only as effective as the data it processes. However, even accurate data can lead to flawed outcomes if it no longer reflects current realities. This phenomenon, known as data drift, underscores the need for ongoing assessment of data inputs to ensure AI models remain relevant and reliable.
We closely monitor evolving regulations on AI and data management, aiming to exceed compliance requirements and uphold the highest standards of transparency and ethical practice. Responsible AI use should be an ongoing commitment for banking and every sector.
As AI evolves, so do the associated risks, including bias, financial crime, compliance issues, and cybersecurity threats. Banks, and all businesses, must prioritise ethical practices and strong governance to harness AI’s potential while fostering innovation. This is the only way we can keep the future of this technology human-centred.
