MK sport under-secretary-general of the UN and rector of United Nations University, delivers a speech titled " src="https://www.globaltimes.cn/Portals/0/attachment/2025/2025-02-24/9f4bd9c0-8a59-4b34-8e97-3e03f487f082.jpeg" gt?="" sheng="" shen="" photo:="" 2025.="" 24,="" february="" on="" beijing="" in="" university="" tsinghua="" at="" future?="" the="" and="" development,="" sustainable="" al,="" />

Tshilidzi Marwala, under-secretary-general of the UN and rector of United Nations University, delivers a speech titled "Al, Sustainable Development, and the Future" at Tsinghua University in Beijing on February 24, 2025. Photo: Shen Sheng/GT

 
Anything that impedes the exchange of ideas and hinders collaboration between countries is not helpful to solve current major global challenges, said a senior UN official in response to an inquiry over the impact of certain countries imposing restrictions on China's AI development and cooperation.

"Discrimination will stifle innovation. If we restrict the flow of ideas between countries, the global challenges we face will become even harder to solve as nations would be working in isolation rather than collectively," Tshilidzi Marwala, under-secretary-general of the UN and rector of United Nations University, told the Global Times on Monday after delivering a speech at Tsinghua University about the future directions for AI advancements and recommendations for AI governance.

Marwala began his speech, titled "Al, Sustainable Development, and the Future," with a joke. He said that first he asked ChatGPT, "What is the relationship between Tshianeo Marwala and Tshilidzi Marwala?" ChatGPT told him "Tshianeo Marwala is the wife of Professor Tshilidzi Marwala," even though Tshianeo Marwala was his grandmother. 

Marwala said the example shows that while AI seems to provide an accurate answer based on possibility prediction, that answer might not be correct. As a result, he proposed a policy recommendation for the future development that "AI must always provide confidence intervals unless the application is time-critical."

Marwala emphasized that before implementing any regulations, it is essential to first identify the core values of AI based on the Universal Declaration of Human Rights and the UN Charter. He described AI values as a balancing act, which requires careful and thoughtful regulation to ensure they align with global principles of justice and equity.

With regard to measures for effective AI governance, he emphasized the need to develop mechanisms for AI incentives, with the clear goal of maximizing the good use of AI and minimizing risks. Mechanism design, identifying the rules, incentives and disincentives will help us achieve this goal.

He also mentioned creating laws on AI and building institutional structures for AI oversight. Marwala highlighted the importance of involving various stakeholders in AI governance and crafting AI policies and regulations based on principles of multidisciplinary collaboration.

China's AI technology, including the emerging DeepSeek model, is gaining prominence. Unlike proprietary AI models, DeepSeek's open-source approach allows anyone to modify and deploy it without oversight. 

While certain countries are reportedly blocking access to or restricting the use of DeepSeek citing "security concerns," Marwala told the Global Times that the UN aims to bring all nations together to live peacefully.

In response to how China's AI progress and its AI governance proposals benefit Global South countries, including China and South Africa, he emphasized that AI cooperation and tech exchanges facilitate capacity building and the sharing of diverse datasets, both of which are crucial for development. He pointed to himself as an example -  going from South Africa to the UN - demonstrating what is possible when people are allowed to move freely.

He pointed out that the progress the world has achieved cannot continue if the flow of information and technological exchange between the Global South and the Global North, or even within the Global South itself, is blocked.

Marwala also addressed the widely discussed question of whether AI poses a threat to human existence. He said that AI is fundamentally a probabilistic statistical analysis tool, providing possibilities rather than a true understanding of the world. He described AI as a reproduction of memory, essentially a rearrangement of existing information.

Additionally, he pointed out that AI operates based on the maximum likelihood approach, which limits its ability to effectively handle events well. This, he emphasized, highlights a key advantage of humans - the ability to deal with cases that have never been seen before.