A working paper by the Centre of Excellence in Financial Services warns that the deployment of AI in SA’s banking system might perpetuate racial biases, calling for industry players to take steps to prevent this and other ethical challenges brought by the technology.
“Some research explores how AI systems embed and perpetuate racial biases. Beyond the data itself, subjective and unconscious biases in AI reflect deeper societal inequalities, often overlooked in approaches that treat bias as a purely technical problem,” the paper reads.
“These systemic flaws affect automated decision-making systems, even within the financial sector, which, without careful ethical consideration, can replicate and amplify structural racism,” it said.
“Many commentators call for greater auditing and accountability in AI systems to identify and address the biases embedded within them. Ethical data-sharing practices and more rigorous oversight are essential to ensure AI systems are fair and equitable. The paper argues that by exposing and mitigating these hidden biases, AI can be designed to challenge, rather than reinforce, existing societal injustices.”
The paper calls for the industry to develop frameworks to reduce bias in AI algorithms by ensuring training data is representative and inclusive of SA’s diverse demographics.
“Regular bias audits and implementing fairness-centric AI design principles are recommended to mitigate potential discrimination in AI-driven decisions.”
The paper investigated the ethical dimensions of AI in SA’s banking sector, focusing on its transformative potential and associated challenges.
Credit agency S&P last year said it expected testing of generative-AI solutions in the banking sector will accelerate over the next two to five years, while benefits are likely to prove incremental.
Banks worldwide are adopting generative AI, which promises earnings growth, improved decision-making, and better risk management. But S&P has warned that AI also comes with new risks, concerns and costs that banks will have to manage.
One of the risks flagged by S&P is ethical concerns, such as the ability to explain generated content or biases embedded in data. Other AI risks that are particular to the banking sector include security and privacy, as well as issues related to workforce displacement.
Standard Bank, Africa’s largest lender by assets, has said it will explore the deployment of AI in its operations though it’s not looking to be a pioneer in the space, but is positioning itself to be a “fast follower”.
The Centre of Excellence in Financial Services stressed that algorithmic bias is another major concern, with stakeholders worried about the potential for AI to reinforce existing inequalities or introduce new forms of discrimination.
The paper additionally said the impact of AI on employment is a significant issue, as automation in banking could lead to job displacement, raising questions about the future of work in the sector.
“In SA banking, AI integration has become increasingly central to operations, particularly in customer service and financial decision-making. The focus is on how AI can enhance banking processes, improve customer interactions, and streamline financial assessments,” it said.
“However, as AI becomes more entrenched in these operations, it brings a host of ethical and practical concerns that must be carefully navigated.
“Stakeholders in this space, ranging from bank customers and employees to management and regulators, are deeply concerned about several key issues.
“Data privacy is at the forefront, as the vast amounts of personal information handled by AI systems must be protected against breaches and misuse.”
Would you like to comment on this article?
Sign up (it's quick and free) or sign in now.
Please read our Comment Policy before commenting.