Navigating the AI Era with Dr Gary Ang:
A Guide to Safety & Skepticism

What mindsets should Singaporeans adopt to navigate this new era safely?

When we use AI, we tend to see GenAI or AI as a monolithic thing. We think: 'Oh, it works very well, we put something in and get something out.' But we need to go deeper and realise that AI is a range of technologies with many different capabilities.

Sometimes GenAI is good for one thing, agentic AI is good for another, and machine learning is good for something else. When you break it down into its parts, you are less likely to fall prey to people who are hawking AI as the solution to everything.

How should Singaporeans adapt to protect themselves from the rise of scams?

In terms of trust and navigating scams, when we see what is happening and understand the state of how bad actors are weaponising AI, we can maintain a more informed mindset and be more sceptical.

We also just need a sense of detachment. These days, if someone calls you out of the blue, you can just say, 'Okay, bye.' Tell them to approach you through official channels or WhatsApp you, and simply refuse to talk further because it just doesn't make sense.

How is a small nation like Singapore protecting users while fostering AI innovation at the same time?

Innovation and governance are actually the same sides of the coin. Having effective risk management that helps you deploy AI safely and effectively is exactly what helps innovation.

Of course, if you put in place rules that are onerous, vague, or lack actual effects, then regulation becomes a blocker. But there is no difference in what a well-developed AI application looks like from both an innovation and a risk management perspective. As long as Singapore focuses on rules that actually help rather than hinder, it is a good thing.

How are financial institutions safeguarding citizens from bad actors leveraging advanced AI?

Breaking that down into two parts. Harms can come from a firm or institution not using AI properly, which is what governance handles. But AI can also be used by bad actors to attack us through cybersecurity breaches, scams, and fraud. Governance frameworks cannot handle that second part—that requires active defensive efforts by the banks and the government.

How they are related is simple: if your risk management is just a checklist that slows down AI development in a way that is not effective, you are actually going to deprive the very people who need these tools to defend against bad actors using AI. Protecting everyday Singaporeans is a priority, but honestly, there is no silver bullet for fraud or scams. We just have to progress and use the technology well to defend against it. We also have to be aware and more sceptical. Unfortunately, many of these things are really not about tech—they are just about us.

Where is the AI risk management industry heading in the next few years?

Before the generative AI boom in late 2022, AI was highly technical and restricted to isolated teams of data scientists. Because of those limitations, governance frameworks remained high-level and principle-based, while standard model risk frameworks caught the rest.

Today, that approach is obsolete. With generative and agentic AI, these tools are in everyone's hands. The surface area for risk has exploded. We can no longer look at AI risk management through a vague principle-based lens. The industry must pivot to concrete, operational risk management practices.

Governance and innovation are not opposing forces. A proper risk framework does not block progress. Instead, it gives organisations the exact guardrails they need to deploy AI applications safely and with actual confidence.