As South Africa was drafting policy papers and convening consultations on how to govern the technology of the future, something quietly extraordinary was happening in Singapore.
There, Vivian Balakrishnan, the country’s Foreign Minister, was doing something that seemed at once futuristic and strangely inevitable: he was building an artificial intelligence version of himself.
Not a symbolic chatbot for public relations. Not a novelty demonstration for a conference stage. But a functioning digital extension of his mind—an AI system designed to communicate on his behalf, manage information flows, and preserve continuity of thought even when he was not physically present.
It was designed to handle WhatsApp and Gmail, reading and sending messages, processing voice notes and images, preparing scheduled briefings, and interacting through a web portal for longer, more reflective conversations. It was developed to research issues, answer questions, draft speeches, condensed complex information, and produced daily updates. It was, by Balakrishnan’s own account, not an experiment but part of his daily operating system—a constantly learning assistant building what engineers would call a knowledge graph, and what diplomats might simply call institutional memory. There is something quietly radical about this.
The person building this system was not Singapore’s Minister of Technology. He was the Foreign Minister.
That fact deserves more attention than it has received.
Because it reveals a truth that many governments are only beginning to understand: in the age of AI, technological fluency is no longer the responsibility of one department. It is becoming a prerequisite for leadership itself.
The old model assumed technology was a sector, like agriculture or transport, to be supervised by specialists while political leaders focused on strategy. That distinction is collapsing. Today, foreign policy is shaped by semiconductor supply chains. National security depends on data governance. Economic competitiveness is inseparable from compute power and talent pipelines. AI is not a department. It is infrastructure.
Singapore seems to understand this instinctively.
Balakrishnan was not alone in developing something impactful, building his digital twin. Josephine Teo, the Minister for Digital Development and Information, was overseeing something equally consequential. In May 2025, she announced the development of MERaLiON—Multimodal Empathetic Reasoning and Learning in One Network—a large language model intended to strengthen Southeast Asia’s AI capabilities.
The project, developed with the Infocomm Media Development Authority, was not simply another race to produce a larger model with more parameters. Its focus was subtler and, arguably, more important: regional language understanding and empathy.
This mattered because most frontier models are trained through the assumptions of distant geographies. Their cultural instincts are often American, their linguistic biases Western, their defaults shaped by contexts far removed from Jakarta, Kuala Lumpur, or Singapore. MERaLiON represented an attempt to correct that imbalance—to ensure that intelligence embedded in machines reflected the people those machines were meant to serve. This was not merely a technical project. It was a sovereignty project.
Together, these two stories—one minister building an AI version of himself, another building a regional language model—tell us something larger about governance. They show what becomes possible when people with relevant skills occupy positions of consequence.
They also illuminate, by contrast, a troubling reality elsewhere.
Which brings us back to South Africa.
Recently, Solly Malatsi acknowledged that the country’s draft AI policy had been based, astonishingly, on fake—or more precisely, non-existent—documents.
It was an embarrassing moment, but embarrassment was the least important part of the story. What it exposed was something deeper: a policy process attempting to regulate a technology it did not sufficiently understand.
One wonders how such a failure could pass unnoticed before being presented to the public. How does a national strategy emerge from documents that do not exist? The answer, perhaps, lies not in bureaucracy but in culture—a culture where performance can sometimes outrun substance, where the appearance of engagement substitutes for actual comprehension.
This is the central dilemma of the AI era: how does one govern what one does not understand?
History offers a useful parallel. In the early days of the internet, many policymakers treated the web as a communications tool rather than an economic architecture. The consequences of that misunderstanding still shape the concentration of digital power today. AI presents an even larger challenge. It touches labour markets, military systems, healthcare, education, industrial policy, and democratic legitimacy itself.
To misunderstand AI is not merely to miss an opportunity. It is to risk strategic irrelevance. That is why South Africa cannot afford superficial leadership at this moment.
AI policy cannot be built through ceremonial participation, keynote speeches, and carefully staged technology summits alone. It requires people in the room who have interrogated models, built systems, understood failure modes, and wrestled with the messy realities of implementation. It requires leaders who know that regulation without capability becomes dependency.
More importantly, leadership in a ministry like Communications and Digital Technologies should belong to someone less interested in appearing at conferences hosted by Big Tech and more focused on building local technological capacity from the ground up.
The task is not simply to regulate imported systems. It is to create the conditions for domestic competence—local models, African datasets, sovereign infrastructure, serious research institutions, and policy frameworks grounded in technical reality rather than borrowed jargon.
South Africa should not waste this moment. The country does not merely need an AI policy. It needs an AI leader.
Someone with a deep understanding of both technology and diplomacy. Someone who recognises that AI is as much a geopolitical question as it is a technical one. Someone who understands that sovereignty in the twenty-first century will be measured not only by borders and minerals, but by who owns the intelligence layer of society.
Such a leader would spend less time delivering keynote addresses and more time asking difficult questions about compute, talent, energy, and institutional memory. They would see AI not as a fashionable talking point, but as a national capability to be built patiently and strategically. Because policy, like software, reflects the quality of those who design it. And nations, like systems, eventually reveal the intelligence of their architecture. The question for South Africa is whether it intends to merely consume the future—or help invent it.
