AI must serve people, not diminish human agency

Artificial intelligence is no longer waiting in the wings as a promising new technology. It is becoming part of the infrastructure through which modern economies produce knowledge, deliver services, communicate with citizens and compete in the global economy. Journalism is already being reshaped in profound ways, from reporting and translation to fact-checking, distribution and the way audiences discover information.

For South Africa, the important question is no longer whether AI will change society. It will. The real question is what kind of change.

We want an AI transition that expands opportunity rather than deepening inequality, that supports development rather than simply increasing efficiency and that remains accountable to people.

That is why South Africa increasingly sees AI as more than a technology issue. It is also a development, an industrial policy, a labour-market and a governance question. Perhaps most importantly, it is a question about the health of the information environment on which any democratic society depends.

Our starting point is our Constitution. Technology, ultimately, must serve human dignity, equality and freedom and contribute to better lives. That requires us to look beyond how quickly AI is being adopted. We also have to ask who is building these systems, who controls the data behind them, whose languages and knowledge are reflected, who benefits from the value they create and who is responsible when an automated system gets an important decision wrong.

These questions will determine what the AI economy looks like in countries such as South Africa.

For us, the first principle is that machines should expand human capability, not diminish human agency. AI can help people work faster and make better use of information, but there are decisions where responsibility cannot be handed over to an algorithm. A system can make recommendations, but it cannot bear democratic accountability.

The second principle is that AI must contribute to economic development. For developing countries, productivity is not enough. We should also be asking whether AI is helping small businesses grow, whether it is creating new industries, whether workers are gaining valuable skills and whether young people are finding new opportunities.

For South Africa, the ambition should go beyond consuming AI products to developing the capacity to produce and innovate with AI.

That requires governance as well as innovation. Regulation is sometimes portrayed as an obstacle to technological progress, but the opposite can also be true. Transparency, fairness, privacy, security and human oversight are not enemies of innovation. They are part of the foundation on which lasting innovation is built. People will trust powerful technologies only when they know there are rules around them.

Data is central to this discussion. AI cannot function without vast amounts of it, which makes questions of privacy, ownership, security and access impossible to ignore. Developing countries need the capacity to protect their citizens' data while ensuring that data can also support research, innovation and economic development.

Then there is language. South Africa has 12 official languages, including South African Sign Language. Yet many AI systems work far better in languages that already have enormous digital datasets. That imbalance may look like a technical limitation, but it can become a social one. If a technology understands some people better than others, it risks reproducing inequality through a new medium.

The possibilities are also enormous. A community radio journalist could report an important public-interest story in an African language and use AI to translate it quickly for a much wider audience. Better speech recognition and translation could make information more accessible in education, healthcare, emergency response and public services. This is what an inclusive AI revolution should look like: technology helping people participate more fully in society without requiring them to give up their language or identity.

But the same tools can also make deception cheaper and faster.

Generative AI has made it possible to create convincing fake images, voices and videos at a scale that traditional verification systems were never built to handle. Imagine a fabricated video appearing days before an election in which a presidential candidate seems to make a racist remark. The campaign denies it. The public argues about it. The video spreads faster than anyone can verify it. By the time it is exposed as a deepfake, the damage may already have been done.

The challenge is therefore not to choose between innovation and oversight. It is to make innovation responsible. That means stronger cooperation on content authentication, deepfake detection, fact-checking, media literacy and the governance of synthetic media.

Journalism itself also needs protection. AI systems are gaining enormous value from news reports, books, films, photographs and other human-created works. African media organizations should not become unpaid suppliers of the raw material for a global AI economy while seeing little of the value created from it. Intellectual property must be respected, creators must receive fair compensation and societies must continue to have access to credible public-interest journalism.

This is an area where China-Africa cooperation could move from discussion to practical results.

The focus should be on building capacity together: developing high-quality African-language datasets with Chinese research institutions and technology companies; training people who can design, audit and govern AI rather than merely use it; improving access to computing power and digital infrastructure; working together on deepfake detection and media literacy; and supporting African companies in creating their own technologies and intellectual property.

Technology sovereignty does not mean isolation. Africa should seek interdependence without falling into dependency. We should be partners in building the technologies of the future, not simply consumers of technologies built elsewhere.

The standard by which we judge AI should therefore be broader than speed or productivity. Does it widen opportunities? Does it reduce inequality? Does it strengthen public trust? Does it respect language and culture? And does it give developing countries a genuine voice in shaping the technologies that will define their future?

South Africa is optimistic about AI, but optimism should not mean surrendering our judgment to technology. We should remain open to innovation while insisting on accountability, and open to international cooperation while protecting African agency.

This matters especially in journalism. AI can assist reporters, but it cannot take on the human responsibility at the heart of journalism: finding the truth, testing claims, exercising judgment and serving the public.

The next chapter of the digital revolution should not simply be written for Africa. Africa must help write it. Its languages, knowledge, people and innovators must be part of the story — not an afterthought.

The author is the deputy minister in the Presidency of South Africa.

The views do not necessarily reflect those of China Daily.

If you have a specific expertise, or would like to share your thought about our stories, then send us your writings at opinion@chinadaily.com.cn, and comment@chinadaily.com.cn.