Remarks by Minister Josephine Teo at the panel discussion 'Harnessing AI To Advance Development' at Semafor's The Next 3 Billion Summit

Q: Singapore was really one of the earlier adopters of AI automation. I was wondering if you could take us through what's working with AI now but more importantly, what's not working? Trying to separate the hype from reality a little for us.

Minister: Well, AI isn't entirely new. We've been experimenting with it in earnest, from about seven years ago. Along with that effort, we also put in place a governance framework. But I should say that the efforts have really taken off in the last three years, particularly with generative AI, making access available to people whose skill levels are not as advanced. So that has been a very significant change.

Now to your question of what is working or what is not working, I think, in the initial stages, there was a real concern about the availability of compute. That has turned out to be not the biggest showstopper. Compute has been made available, and ambitions are growing. Right now, there is a very healthy level of interest in enterprise adoption. While it started with the idea of using generative AI for workplace productivity improvements, such as transcribing and summarising, what has been really interesting are the applications we are now seeing in specific verticals.

For example, in manufacturing, we have seen deep interest in how AI can be used. In all manufacturing companies, there is a need for operations planning, defect detection and maintenance. For example, in semiconductor manufacturing, wherever a tool in the production process is put out of action, a lot of money is lost. So, if we can use AI to do predictive maintenance, that can potentially save a lot of money. Therefore, those kinds of applications are gaining a lot of traction.

What's really interesting is also the kind of AI specialists that are needed. We need not only people who are skilled in data science and machine learning, but also those who understand the actual domain - the functional experts. So, the marriage of domain functional experts together with data scientists and machine-learning experts make a formidable team. In fact, a person who has both sets of skills is what we call an "AI bilingualist", and they are even more valued by companies.

Q: Minister Teo, I'd love to get your point of view. I think Singapore has been right in the middle of this, with the chip issue allegedly being diverted to China. But then the US policies are going back and forth right on technology diffusion. How are you navigating that sort of whipsaw effect in US policies?

Minister: We do not condone illegal activities taking place either within Singapore or through Singapore. Our position is very clear. If there is any evidence that we should follow up, we will. I think our US colleagues know it. To the international community, Singapore has been very consistent.

I should also say that we maintain an open economy. Organisations do have a choice in the way AI is being adopted. They will have to go with the models that best serve their interests. The decision they make could be a combination of performance, as well as the security and resilience of the technology, balanced against cost considerations. These are evergreen issues that companies implementing all kinds of technology projects will always have to consider. They should be able to make a considered decision, balancing the different trade-offs. That's the position we take.

But having said that, I think it is fair to say that geopolitical issues do weigh on organisations - they will be thinking if a technology stack is unsafe for their users, presents security issues, or risks becoming unreliable or unavailable. Organisations implementing the technology will have to weigh these considerations and assess if the risk is too big for them to proceed.

For Singapore, we would like as best as possible to support countries in each of our endeavours. So Singapore's Large Language Model, which Dr (Bosun) Tijani talked about, is developed to reflect the richness and diversity of the language environment in Southeast Asia, In time to come, if other regions were to develop similar models, there will be scope for us to work together to overcome the challenges of AI models in reflecting our cultures and values. This is an important endeavour, and we all want to see AI models reflecting our identity.

Q: Do you think you'll also benefit from being sort of neutral? I noticed like it was interesting to see Manus AI move to Singapore. Because they didn't want to be "Chinese-based" anymore.

Minister: Well, organisations have to assess whether Singapore provides the kind of environment that enables them to do their best work. This means whether the infrastructure is available, whether the skilled workforce is available, and very importantly, whether there is respect for intellectual property.

It is not just a case of neutrality but there are many factors that a company must consider before deciding where to locate their most important activities. So, we would like to believe that it is not just geopolitical considerations, but the inherent fundamentals that make Singapore attractive for companies to locate their key activities where we are.