Generative AI in Finance: Navigating Innovation and Trust in the Digital Age

The recent Personal Data Protection (PDP) Week 2024 underscored the growing significance of generative AI in the financial sector. While this new technology presents transformative potential, it also has raised critical concerns regarding data privacy and trust.

Generative AI Adoption In The Financial Sector

Generative AI is already changing the way people work. OCBC Bank was the first local bank to go big on generative AI, putting a chatbot at the command of its 30,000 staff across 19 countries. This was implemented after 1000 members of their staff underwent a trial run of their internal generative AI app and employees feedback that they were able to complete their tasks in half the usual time it takes. OCBC is now piloting four generative AI functions for their employees. The bank’s AI bot is expected to make up to 4 million decisions on a day on risk management, customer service and sales for the bank.

The Data Dilemma

Generative AI models, such as OpenAI's GPT series, rely on large amounts of data, often containing sensitive financial information. This has raised valid concerns about potential misuse or breaches especially among the financial community. 

As Minister Josephine Teo highlighted, acquiring high-quality data for AI development is challenging. Even when accessible, these data may contain Personally Identifiable Information (PII), leading to biased outcomes or privacy violations. The scarcity of quality data could become a limiting factor for these models as early as 2026.

Singapore's Proactive Measures

Singapore is taking a proactive approach to address these challenges through a multi-pronged strategy:

        1. Model AI Governance Framework: This comprehensive framework fosters a trusted AI ecosystem by emphasising accountability, transparency, and rigorous testing throughout the AI development lifecycle. It also provides guidance on clarifying accountability in the AI value chain, improving the model development lifecycle, and addressing cybersecurity and misinformation risks.

        2. AI Verify Framework: AI Verify’s AI Governance testing framework and software toolkit introduces safety guidelines for generative AI model developers and app deployers, prioritising transparency and robust testing protocols. Developers are encouraged to disclose information about the data used, testing results, as well as potential risks and limitations to ensure the models are safe and trustworthy.

        3. Privacy Enhancing Technologies (PETs): PETs, such as synthetic data generation, are being promoted to mitigate risks associated with using sensitive financial data. The Personal Data Protection Commission (PDPC) has released a "Proposed Guide on Synthetic Data Generation" to help businesses understand and implement these technologies effectively.

Regional Collaboration

Singapore is actively collaborating with ASEAN countries to harmonise standards and share best practices in Data Governance. This includes the upcoming release of the ASEAN Guide on Data Anonymisation, which will provide guidance to businesses in the region on anonymising data for greater and more responsible use.

Implications for Financial Professionals

For risk management professionals, data protection officers, legal counsels, and IT professionals in the financial sector, these developments highlight the importance of staying informed and adapting to changing tides. Understanding the intricacies of AI governance frameworks, the potential of PETs, and the evolving regulatory landscape is crucial for navigating the complexities of generative AI in finance.

The Road Ahead

Generative AI today is akin to the internet circa 1996. It stands impressive relative to prior technologies but pales in comparison to the advancements anticipated in the next 15 to 20 years. The history of innovation is compelling, often showcasing how many companies fail to recognise and seize opportunities, subsequently being outpaced by newer, more agile competitors.

Certain companies are strategically positioned to capitalise on the generative AI wave, leveraging its potential to their advantage. Conversely, others will falter and fall behind. The financial industry in Singapore has kept pace with generative AI, witnessing transformation from personalised customer service to sophisticated fraud detection. 

However, building and maintaining trust is paramount. By adopting a balanced approach that prioritises both innovation and data protection, the financial sector can harness the full potential of AI while safeguarding sensitive information.

We’ll be sharing more about this at our upcoming industry talk, “The Good, The Bad & The Ugly - Risk Management In The Age Of AI” on 16 August 2024. Andeed Ma, President of the Risk and Insurance Management Association of Singapore (RIMAS), partner of the Artificial Intelligence International Institute and author of AI for Humanity: Building a Sustainable AI for the Future, will be joining us at the session to speak on AI risks and mitigation strategies. Secure your spots here.