How to use AI responsibly

Summary

Generative AI (GenAI) promises revolutionary impacts on finance, from automating loan approvals to personalized investment advice. However, ensuring its benefits extend to all users necessitates diverse data training. This blogpost explores GenAI’s potential and challenges like data diversity, illustrating initiatives like Singapore’s Project MindForge aimed at fostering responsible GenAI use.

The Promise of Generative AI in Finance

Imagine a future where robots manage your finances instead of people. This is the potential of Generative AI (GenAI) in the financial sector. GenAI refers to AI systems that can generate new content and solutions after learning from vast amounts of data, from making investment suggestions to evaluating loan applications.

However, there’s a significant hurdle: what if GenAI systems fail to serve a segment of society, simply because they haven’t been trained on diverse enough data? Consider a scenario where an AI system denies loan applications for small business owners in rural areas more frequently than those in urban centers, simply because it has been predominantly trained on data in the urban areas. 

This challenge underscores the importance of data diversity in AI’s development and application. By “data diversity,” we mean using a broad range of data sources that represent different demographics, geographic locations, and economic backgrounds. Ensuring that GenAI systems are trained on diverse datasets is crucial to developing solutions that are fair and effective for all users, reflecting the diverse clientele they serve.

By addressing these concerns, we can unlock the full potential of GenAI to revolutionize financial services, making them more inclusive and accessible to everyone.

The Challenge of Data Diversity

The potential of GenAI in finance is immense. According to the McKinsey Global Institute, the banking industry stands to benefit significantly from GenAI, with projections suggesting an annual increase in value ranging from $200 billion to $340 billion. The use of GenAI spans various areas, from enhancing customer service to complex tasks like credit scoring and fraud detection. However, its application must be carefully considered, particularly in sensitive areas like loan approvals, to avoid biases and maintain fairness. This introduces the challenge of data diversity, a crucial factor in responsibly deploying GenAI technologies. Recognizing these challenges, forward-thinking initiatives have begun to emerge, aiming to address these very concerns.

Singapore’s Project MindForge: A Proactive Approach

Singapore has taken a proactive stance in managing the risks of GenAI. Led by Minister for Communications and Information, Josephine Teo, and introduced at the World Economic Forum, this initiative is a testament to the nation’s commitment to responsible GenAI use in finance. Project MindForge seeks to develop a comprehensive risk framework that addresses critical areas like accountability, governance, transparency, fairness, legality, ethics, and cybersecurity. Companies involved in Phase 1 include DBS Bank, OCBC Bank and United Overseas Bank Limited.

Real-world Applications in Finance

OCBC’s adoption of GenAI to enhance customer experience is a prime example of the technology’s vast potential. This goes beyond customer interactions, extending to areas such as automated loan approvals and tailored investment advice. An IMF paper warns of inherent risks in GenAl technology, including embedded bias, privacy concerns, outcome opaqueness, and lack of performance robustness. To gain public trust, financial institutions must emphasize user privacy, allowing customers to control their data sharing preferences. Moreover, the decision-making process in AI should involve significant human oversight to ensure ethical and responsible use of technology. 

Another significant application within Singapore’s fintech sector is PolicyPal’s innovative use of AI in insurance. This startup is allowing personalise insurance coverage by helping individuals  manage and optimize their insurance coverage. By employing AI, PolicyPal offers a more intuitive and personalized insurance experience, showcasing how technology can make complex financial services more accessible and user-friendly. This approach not only makes insurance more accessible and relevant to a diverse range of clients but also represents a significant leap from traditional, one-size-fits-all insurance models. PolicyPal’s success in utilizing GenAI showcases how such technology can transform a sector as intricate as insurance, making it more efficient, customer-centric, and inclusive.

Beyond its current applications, the potential of GenAI to revolutionize the financial sector extends to areas yet to be fully explored. One such area is the development of AI-driven financial advisors that cater specifically to the needs of senior citizens, offering simplified interfaces and tailored advice. This innovation could bridge the gap in financial literacy among older generations, making complex financial information more accessible and understandable. 

Recent incidents, such as the 2022 attack on Revolut, where the hackers had access to the details of about 32,000 customers for a short duration, underscores the critical need for GenAI in enhancing cybersecurity measures. GenAI enables financial institutions to anticipate and mitigate vulnerabilities, thus ensuring a safer banking environment. By analyzing patterns and predicting threats, GenAI strengthens defenses against exploits, safeguarding financial transactions and enhancing consumer well-being.

Disclaimer: The views and opinions expressed in this article are solely those of the author and do not reflect the official policy or position of the National University of Singapore (NUS) or the NUS FinTech Lab.