Speech by Minister Josephine Teo at NTUC LearningHub Human+ Symposium 2026

Good morning and thank you for inviting me to the Human+ Symposium.

For many years now, NTUC LearningHub has been a strong partner in training and upskilling our people - working with the Government, employers, and TACs.

I am glad that you are championing the Human+ initiative to enable Singaporeans to succeed in the age of AI.

1. _p]:inline">

   This certainly aligns with the Government's efforts in recent years, to systematically develop AI capabilities in our enterprises and workforce.

Many countries and organisations are taking note of our initiatives, as they consider their own responses to the present challenges and opportunities.

1. _p]:inline">

   This time last week, I was still in Davos attending the World Economic Forum.

2. _p]:inline">

   Of course, unsurprisingly, geopolitical tensions took centre stage.

3. _p]:inline">

   But I should share with you that there was continued strong interest in AI developments and Singapore's pragmatic approach, including how we equip our people to stay relevant.

We keep a close watch on emerging trends and current assessments of AI's impact.

One such discussion caught my attention. It involved the founders of Google DeepMind Demis Hassabis (a 2024 Nobel Prize Winner for Chemistry) and the founder of a company Anthropic Dario Amodei. There's a YouTube video of their exchange, in case you're interested.

1. _p]:inline">

   These two founder-CEOs had debated how AI could affect societies.

2. _p]:inline">

   They spoke on the importance of economies and labour markets remaining adaptable.

Demis was in Singapore a little over a year ago and met Prime Minister Lawrence Wong. At Davos, he updated me that DeepMind's presence in Singapore was growing well. They have recruited key members of their team. It includes a Singaporean, Dr Yi Tay, who helps to train the Gemini models, your organisations may be using.

I also caught up with Dario, and the conversation with him crystallised something for me.

1. _p]:inline">

   He told me that he spent his year-end vacation writing a 20,000-word essay on the challenges brought about by AI, titled "The Adolescence of Technology".

2. _p]:inline">

   This is a follow-up to his previous essay "Machines of Loving Grace", which focused on the upsides of powerful AI for humanity.

3. _p]:inline">

   In this recent essay, Dario deals with the possible dark sides of AI, the risks, and our possible defences.

4. _p]:inline">

   He suggests that one way to understand the scale and speed of upcoming AI progress is to imagine that a "country of geniuses" materialises somewhere in the world next year. Except that this "country" is contained in a data centre, literally.

5. _p]:inline">

   Everyone in this "country" of around 50 million would be more capable than any Nobel Prize winner, statesman, or technologist, and operate much faster than humans.

6. _p]:inline">

   I think, such a future, even if described only in words, inspires both hope and fear.

7. _p]:inline">

   I would encourage you to read both essays, or minimally to get Claude or ChatGPT to summarise them for you.

The debate on the imminence and risks of very powerful AI will continue a while more.

1. _p]:inline">

   But one thing is already clear.

2. _p]:inline">

   We must find better answers to questions about AI safety. We must also learn how to develop and deploy AI in responsible ways.

3. _p]:inline">

   I have on previous occasions spoken on Singapore's efforts on this front.

4. _p]:inline">

   Most recently, last Saturday evening, I launched the National AI R&D Plan, which will support research in fundamental issues in AI, including 'Responsible AI'.

There is also urgency for us to grow new workforce capabilities, and become adept at partnering with "country of geniuses in a datacentre".

I have spoken in the past about the potential of AI to amplify human capabilities, acting as our 'teammate'.

1. _p]:inline">

   We want to shape a future of work that is not about humans versus AI, but about humans doing better with AI.

This is especially relevant when the centre of gravity has shifted from extracting productivity improvements through individual use of AI, to the prospect of contingents of AI agents carrying out multiple tasks on behalf of humans.

Naturally, the role of humans will evolve.

There are already examples from the field of software development. You must have heard of them.

1. _p]:inline">

   Many software developers today no longer write long chunks of code.

2. _p]:inline">

   Instead, human software developers direct coding assistants with clear problem statements, validate output from agents with critical thinking, and use design principles to orchestrate systems.

3. _p]:inline">

   It's like a conductor standing in front of musicians, telling them "Here is where you increase the tempo, here is where you soften the tone, so that the music comes across as soothing and melodious."

4. _p]:inline">

   These, of course, require a deeper domain understanding. Just as a conductor does not necessarily play any particular instrument very well, he needs to understand the principles of music and how the sounds come together, Developers will need to be able to collaborate effectively with AI agents and systems too. These are fields that we are only beginning to discover, and have to figure a way out to make them work in the future.

Such developments lead to the question of how we should think of humans working effectively with AI?

It may not always be possible, but as a starting point, our orientation should be to use AI to augment our capabilities, rather than replace us.

One way of achieving this is becoming fluent in working with AI systems - just like learning how to communicate effectively in a specialised domain.

1. _p]:inline">

   What we bring to the table is our "mother tongue" - our domain knowledge, what we know and can do that remains challenging for AI.

2. _p]:inline">

   We can point to several types of "mother tongues" or domain expertise - HR, manufacturing, law, and public policy. Being good in our mother tongue is a baseline requirement.

3. _p]:inline">

   The new requirement is developing familiarity or fluency in AI collaboration - understanding how to direct AI systems, interpret their outputs, and integrating their capabilities into our work.

Becoming bilingual in this sense helps us achieve our aim of doing better with AI, instead of competing against it.

1. _p]:inline">

   The value comes from being able to use fluency in AI and domains to transform processes and improve outcomes.

2. _p]:inline">

   For example, in manufacturing, a process engineer who knows enough data science and machine learning, can use them to increase production yields and reduce material wastage.

This effort of acquiring new fluency is not easy but neither is it impossible. It requires learning and re-learning, a practice that is, fortunately, not entirely new to Singaporeans.

1. _p]:inline">

   The entire SkillsFuture movement is built in this spirit.

2. _p]:inline">

   Even before that, Singaporeans had adjusted to previous waves of technology-led changes to work - computerisation, the internet, mobile.

Being bilingual - with fluency in AI and domains - will similarly help our people work with AI like teammates. This can be a formidable force for Singapore, to remain relevant and competitive as a workforce.

Undoubtedly, this will take considerable effort. But we are working from fairly good foundations.

1. _p]:inline">

   Globally, Singaporeans are one of the highest per capita users of AI platforms like ChatGPT.

2. _p]:inline">

   This extends to the workplace.

3. _p]:inline">

   IMDA's recent Singapore Digital Economy Report showed that three in four workers surveyed are already using AI tools regularly at work.

4. _p]:inline">

   85% reported improvements in productivity and work quality. We know the headroom for further improvements is quite high.

5. _p]:inline">

   We now have more than 60 AI Centres of Excellence uplifting our capabilities across many different sectors.

The Government will continue to help our businesses and people upgrade themselves and utilise AI effectively as a teammate.

NTUC LearningHub's Human+ initiative is very much aligned with our plans, and I thank you for it.