Opening Address by SPS Syed Harun Alhabsyi at NUS-ISS Annual Luncheon 2026

It is truly a privilege to join you today, and to be amongst a community committed to building the skills and capabilities that we need for a fast-changing world.

Each and every one of you knows this already. AI is increasingly a big part of that future, and I think all of us here are intimately familiar with its growing possibilities as well.

However, there are also some harder questions with AI that I’d like to reflect with you today:

1. First, as AI does more for us, what then happens to our own expertise? You may know that I was a clinician before. And sometimes with AI coming to the fore, as a clinician, one wonders what that leaves us with, as far as expertise is concerned.

2. Second, is AI helping us work better, or adding to the pressures of work?

3. Third, how do we make sure the benefits of AI extend to those outside the tech sector, and into other areas and professional fields too?

On the first question, AI is already moving beyond answering questions and generating content. Increasingly, it can plan and act on our behalf. That opens up enormous possibilities, but it also raises the bar for the human capabilities that are needed alongside AI.

Earlier this year, both OpenAI and Anthropic reported troubling incidents while testing their models’ cybersecurity capabilities.

OpenAI’s models found ways around safeguards and accessed real systems, while Anthropic’s models, mistakenly given internet access, broke into systems of real organisations, believing they were part of the exercise.

These incidents illustrate a broader challenge. As AI systems become more capable, they may behave in ways that even companies at the frontier of AI cannot fully anticipate. In response, these companies are now investing heavily in better understanding their models, and learning how to keep them in check.

This challenge is not just a technical one, nor is it just about keeping the human in the loop. The more responsibilities we give AI, the more important it becomes for us to retain the expertise to direct AI, to question AI, and certainly to recognise and step in when something goes wrong with AI.

Additionally, as we rely more and more on AI, might the very skills we need to exercise that judgement be eroded over time? It is something we must consciously guard against.          

McKinsey recently alluded to this, too. In a recent study, they found that tasks such as research, documentation, basic coding and preliminary analysis are increasingly being streamlined by AI. Yet, these are also the very tasks through which junior employees have traditionally built instincts and judgement.

If AI saves someone hours on a task, that is a real benefit. But we also learn by working through difficult problems, making mistakes and seeing why something went wrong. If AI increasingly does such thinking for us, we should consider how future generations will hone these capabilities in turn.

The answer surely cannot be to preserve old ways of working for their own sake. It is to find new ways to build expertise. AI itself could be part of this too, by challenging our reasoning, offering different perspectives and helping us practise situations we might not otherwise encounter.

This matters especially as skills like judgement, creativity and our sense of empathy become more important in an AI-enabled world. These capabilities do not develop automatically. As we invest in more capable AI, we need to be equally deliberate about developing more capable people in our mix.

Second, we are still learning what it means to use AI well at work and to work better.

The promise is often that AI will save us time. But there is also an emerging tension: what really happens to the time we saved?

Earlier this year, UC Berkeley researchers studied how generative AI has changed the work at a tech company. They found that workers now moved faster, took on a wider range of tasks and extended work into more hours of the day, often without being asked. AI had made more possible, but work had also become more intense.

This distinction matters. Productivity gives us the capacity to do better work, not just fill every hour with more work.

We should also acknowledge that using AI well actually takes a lot of effort. We have to learn new tools, work out where they could be useful. We have to check their output and adapt our ways of thinking about work. And just when we become comfortable, the tech moves again, and we have to learn again.

For an already busy employee, and the employers will know this, learning can easily become yet one more thing to fit into the working day. If we want people to become fluent in AI, we cannot simply hand them the tools and then expect them to figure it out.

It would require a broader understanding of AI fluency. Beyond knowing how to prompt or use the latest app, it lies in recognising where AI genuinely improves our work, where it may not do so, and how to use AI with confidence and good judgement.

In this regard, the many employers here in the room, you play a key and vital role. Workers do need the time and support to experiment, to learn from one another, and to apply AI in their work with confidence. Training must also be grounded in the realities of their jobs, with opportunities to put what they have learnt into practice.

We are trying to support this in Singapore too. Since September, eligible Singaporeans who attended selected AI courses have been given complimentary access to AI tools for six months, so that learning does not stop when the course ends.

This is also where institutions like NUS-ISS can make a real difference: working closely with industry to understand the problems people face at work, and help them build the skills to address these problems, even as their job tasks evolve.

Ultimately, learning should give workers greater confidence and opportunity to do better, and not become another pressure they have to carry.

Finally, the benefits of AI cannot be confined to the tech sector alone. Some of the greatest opportunities from AI will come from putting AI in the hands of people who already possess deep expertise in their fields.

We are not saying that lawyers, doctors and teachers must concurrently become AI engineers. But the professionals of the future will need to appreciate where AI can help in their respective domains, where its output needs to be checked, and where professional judgement still matters.

This is the idea behind Singapore’s push for workers to become AI bilingual – combining expertise in their respective fields, with the ability to use AI effectively. Under the National AI Impact Programme, we have set out the target and goal to train 100,000 non-tech professionals to become AI bilingual over the next 3 years.

We are already starting to put this into practice. In July, IMDA and the Institute of Singapore Chartered Accountants launched “AI x Accountancy”, the first AI fluency programme for non-tech professionals under the National AI Impact Programme. It is designed around the actual work of accountancy and finance professionals, including applying AI to financial analysis and other role-specific workflows.

This illustrates another important point. AI fluency will look different for different professionals. What an accountant needs from AI will surely be different from what a healthcare professional, doctor, lawyer or educator will require.

The people who understand these professions best need to help shape how AI is used in their work. This means bringing domain expertise, technology and training together.

This matters especially for smaller organisations, our SMEs, who may understand their businesses operations and models well but lack the resource and expertise to explore AI on their own. Conversely, a tech provider, many of you here, may know the tools very well, but not necessarily the realities of the business and how AI can effectively translate to real impact and outcomes for the businesses we want to be able to assist.

Our ambition must therefore go beyond just helping people learn about AI. Partnerships that bring together different strengths can help solve real problems with AI, and even rethink how work is done.

Earlier, I got the chance to meet some of the award winners, and I was glad to see NUS-ISS taking this approach through the Show Me Your Agents Hackathon, bringing together SMEs, students, professionals, and technologists to develop agentic prototypes around real business problems.

1. A team I met earlier – they call themselves Team Interns – but the amount of work they had to do was hardly within the remit of interns alone. They were students in design and AI from Singapore Poly, Temasek Poly and NTU, and they work alongside an employee from a local contracting firm.

2. Together, they applied their combined expertise to a problem familiar to many businesses: customer enquiries, invoices and service requests, scattered across different systems.

This shines the light of possibilities, when different strengths and perspectives come together to unlock more of AI’s potential for all.

We have good reason to be optimistic about AI.

At the same time, it is worth being thoughtful about the choices we make as AI increasingly becomes part of our everyday work.

1. First, this has to include continuing to develop the depth of judgement and expertise to guide an increasingly capable technology in our midst.

2. Second, this means making learning more practical and supportive for workers and employers, so that they can benefit from AI without feeling that they have to navigate these changes alone.

3. And third, it also means building partnerships that bring AI capabilities together with the knowledge and experience that people already have.

These are things that educators, employers, industry partners, and Government can continue to shape together. I am grateful to NUS-ISS, and to everyone here, for the work you are doing.

Ultimately, the promise of AI is not simply that machines will become more capable. It is that people and our societies will become more capable too, with more time and capacity for what truly matters.

And that certainly is a future worth working towards.

Thank you.