The United Nations has launched its first global panel on artificial intelligence, as concerns grow that the technology could deepen global inequalities – particularly in Africa, where systems are largely imported after being shaped elsewhere.
The panel, bringing together around 40 experts from 37 countries, was approved by the UN General Assembly in February and held its first meeting in March. Members serve in their personal capacity for a three-year term.
It aims to help governments make sense of artificial intelligence (AI) as its reach quickly spreads across economies, politics and everyday life – and to close what the UN calls a growing “knowledge gap” around the technology.
Comparable with the IPCC climate change panel, it is designed to provide independent scientific advice and produce regular assessments of AI’s risks and impacts, at a time when a handful of companies, mostly in the United States and China, dominate the field.
Among its members is Senegalese researcher Adji Bousso Dieng, who tells RFI that Africa needs to develop its own AI or risk being left dependent on others.
RFI: What is this new UN panel on AI meant to achieve?
Adji Bousso Dieng: AI is advancing at an unprecedented speed and is now entering many parts of our societies – our economies, science, politics and even culture. Many governments and decision-makers feel uncertain.
They see the huge potential of AI, but they still struggle to fully understand its implications, how to use it for the common good and how to protect against its risks. That's why the UN created an independent scientific space. We do not work for any government or institution.
Our goal is to produce rigorous scientific analysis to guide public decisions. In the end, it is about rebalancing things so that AI governance and access to opportunities are not concentrated in the hands of a few actors, but benefit the whole international community.
EU votes to ban AI 'nudifier' apps after explicit deepfake outrage
RFI: AI is clearly dominated by major companies, especially in the United States. What's your view on this?
ABD: That is a reality. We keep hearing the same names – OpenAI, Anthropic, and now DeepSeek from China. Fortunately, other start-ups are emerging, including in Europe with Mistral. But it is a problem that AI development is concentrated in the hands of these companies.
AI has become a public good. Countries in the Global South should be able to develop their own AI, but they do not have the means. It requires huge resources in computing power, data and skills. So we need collaboration between nations, and this UN panel is a good way to start that work.
RFI: How is AI developing in Africa?
ABD: There are communities working on AI, like Indaba, and companies are starting to use it. But I am not satisfied with how it is happening. Most systems come from outside, they are not developed locally. That creates problems, especially bias.
Today’s most powerful AI systems are trained mainly on Western data, which does not reflect the diversity of populations. We need local AI systems built with local context, so they can solve local problems. Many systems try to give one single “best” answer, but that can lead to repetitive and biased results.
My research focuses on introducing diversity into AI, so it can explore multiple solutions and hypotheses. That is essential in science, because discovery is about exploring new ideas. We developed a mathematical tool called the 20-10 score to measure and guide this diversity, so AI becomes more exploratory, more creative and closer to scientific thinking.
RFI: Is it possible to develop a pan-African AI?
ABD: Yes, that is one of my goals. In many areas, we need a pan-African approach – but a practical one. Not just political slogans about sovereignty, but real collaboration to solve concrete problems in education, training and trade. We need this cooperation across the continent in all fields, including technology and AI.
RFI: You have spoken about a form of digital colonisation in Africa. What do you mean by that?
ABD: For example, companies go to countries like Kenya to label data, which is needed to train AI systems. The working conditions are often not fair, people are not well paid and they can be exposed to traumatic content.
There is no proper legal framework. That is a form of digital colonisation. There is also the issue of data sovereignty. Data can be used without compensation, and large companies benefit without paying Africans for their work.
RFI: Are people and governments aware of these risks?
ABD: I do not think so. There is a lot of enthusiasm for AI in Africa. People believe it will solve many problems – healthcare, education, jobs. But that is not entirely true. There is work to do to build local AI systems.
Right now, Africa risks repeating what has happened with natural resources – being a consumer rather than a creator. And I do not think this is discussed enough, especially by governments.
AI images 'distort memory' of Holocaust, experts warn, as survivor numbers shrink
RFI: You were born in Senegal, studied in France and now work in the United States. What has guided your journey?
ABD: I have been very lucky to have experiences in Senegal, France and the United States. What has stayed constant is my love of knowledge. I am very curious and passionate about science. The work we are doing at Princeton is something I truly believe in.
AI should not just be a prediction tool for chatbots like ChatGPT or Gemini. It can become a partner in discovery, helping solve major global challenges. That is what motivates me.
RFI: You founded the NGO The Africa I Know to encourage young people, especially girls, to go into science and AI. How does it work?
ABD: The idea is to give young Africans the tools to become creators of technology, not just consumers. We do this through inspiration, with videos about Africans succeeding in AI and other STEM fields. We also run a summer camp.
Students learn both the opportunities and limits of AI, then the basics, and then work in groups on a project to solve a local problem using AI. They are incredibly creative. It is important they build their own technologies, because they understand their communities’ problems better than anyone else.
RFI: Despite the risks, are you optimistic?
ABD: Yes, I am optimistic about using AI for science. Traditional research can take a long time. AI can speed up the discovery of new molecules or materials for energy, climate, health and agriculture.
But there are also risks. Chatbots can be addictive and make people too dependent. There is a danger of losing critical thinking and creativity.
This interview has been adapted from the original version in French and lightly edited for clarity.
