What if participation—not profit—fueled AI development? In this TED Talk, researcher and policy expert Nanjira Sambuli proposes a different way forward for creating AI models, an avenue rooted in African indigenous traditions and values.
Nanjira Sambuli, “What Ancestral Intelligence Can Teach Us about AI.” TED.com, April 2025.
- Sambuli begins her argument about AI development with a metaphor based on an African proverb: “When elephants fight, it’s the grass that suffers” (0:30). In this metaphor, who symbolizes the elephants? Who is represented by the grass? Write 1–2 sentences that summarize her view about which individuals and groups wield power in the race to create more powerful AI technologies, and who (and what) bear the consequences.
- Large language models require a massive amount of data (3:57). Sambuli offers an alternative to this model, which she names “ubuntech” (2:33). Where does this name come from? What’s one way “ubuntech” is different than conventional AI models and companies?
- So what? Who cares? Find where Sambuli says why her argument matters. How do Africans benefit from “ubuntech” AI companies and models? Who, beyond Africans, could benefit from her vision for AI development? What should be the ultimate goal of AI technology, according to Sambuli?
- Some scholars argue for rejecting or “refusing” AI technology, in part due to its broad ecological and labor costs. How might Sambuli respond to this naysayer argument?
- Today’s AI models are trained on only a few languages, with 90% of the data coming from mainstream American English. Read this conversation with Chenai Chair, the director of the Masakhane African Languages Hub, an initiative Sambuli mentions in her video. Why is a lack of linguistic diversity in AI models a problem?
They Say I Blog 
After reading the article about Nanjira Sambuli’s ideas on creating more fair and sustainable AI systems, I found her argument really important, especially her focus on who gets to be involved in building technology. The author argues that AI today is mostly controlled by a small number of powerful companies, which means the systems often reflect their interests instead of representing everyone. Sambuli introduces the idea of “ubuntech,” which focuses on inclusion and making sure more communities have a voice in how AI is developed.
Many people on this blog have made similar points about technology and fairness. For example, in discussions about AI tools in schools, some argue that these systems can be biased and unfair to certain students. In the same way, Sambuli suggests that AI is not neutral and can reinforce inequality if only a few groups are in control. I agree with this because it makes sense that the people creating technology influence how it works and who it benefits.
At the same time, some might argue that large tech companies are necessary for innovation and progress. While I understand that point, I think Sambuli’s argument is stronger because it focuses on long term fairness rather than just short term growth. Overall, this article shows that AI is not just about advancing technology, but also about making sure it is developed in a way that benefits everyone.
LikeLike
As a result of my reading of the article written by Nanjira Sambuli, I share the opinion of the author about the necessity of creating AI in such a way that will benefit everyone and not only several powerful companies. The author argues that creators of AI have an enormous power over the way in which technologies are used. When the creation of AI is controlled by several wealthy nations and big technology companies, the interests of other communities and the problems they face will not be considered. It is essential that different cultures and languages should be involved in this process of creating AI for the future. An interesting idea from Sambuli is her statement on the use of language diversity in AI. According to her, AI systems learn based on data available in a very limited number of languages; more than 90% of data belongs to mainstream American English. It means that users who speak some other languages will not get the right response from AI applications. The problem itself has never occurred to me prior to reading the article. In case AI does not work in different languages and cultures, it cannot help all people. Although I concur with Sambuli entirely, I am convinced that developing ethical AI requires contributions from various entities. The government, educational institutions, tech companies, and consumers all need to contribute to the development of an AI system. Tech companies need to collect more diverse data, while educational institutions are supposed to educate students on the functioning and use of AI. However, the consumers must be aware that AI is a helpful technology, but it still has certain flaws and may demonstrate bias. It means that we should not accept the findings without analyzing the information. In summary, Sambuli’s article proves that the development of AI is not just the improvement of technology, but its development for everybody. I am confident that involvement of different people in AI creation will allow developing future AI systems representing interests of people worldwide.
LikeLike
After I watched Nanjira Sambuli’s talk, one thing made me think more. She focuses on that who controls AI and whose lives become data. Before this talk, I thought data was only numbers and information from the internet. But actually, some data comes from people’s language, culture, and even knowledge. If a company uses these things to train AI and make money, the community definitely lose control of them. People should know that how their information is used and have a chance to deny. I agree with Sambuli’s idea about participation, while participation also needs more than asking people for their opinions and thoughts. A company may invite different people to a meeting, while the company still makes every important decision. Then the people are included, but they do not have real power about that. For example, AI might be used to choose a person for a job interview or decide if someone can get a loan. If AI makes a wrong decision, that person should be able to question it and correct the information. Without this choice, big companies are like elephants and regular people become like the grass under them. Some people may argue that giving communities more control will definitely make AI develop slowly. I can understand this because companies need to compete their task. While, making AI very fast is not always better. A mistake in an entertainment app may not matter, but a mistake in health care, hiring or banking can change someone’s life. Even taking more time in the beginning may avoid a bigger problem later. I also think communities should receive something when their knowledge helps companies earn money, such as jobs, training or new technology. Sambuli made me realize that using AI and having a voice in AI are different. Real participation means people can use the technology, question it, and have some control over it.
LikeLike
I agree with Nanjira Sambuli’s argument that artificial intelligence should be developed according to values of responsibility, participation, and shared humanity rather than only competition and profit. Her use of the proverb “when elephants fight, it’s the grass that suffers” is especially effective because it shows how powerful governments and corporations can compete over AI while ordinary people and vulnerable communities experience the consequences. The metaphor also suggests that technological progress is not automatically beneficial. Progress can become destructive when the people providing data, labor, knowledge, and natural resources are treated as tools instead of participants.
Sambuli’s discussion of ubuntu gives the speech its strongest ethical foundation. The idea that “I am because you are” challenges the belief that technology can be separated from the people and environments it affects. AI systems are often discussed as if data were simply a resource waiting to be collected, but Sambuli reminds us that data represents real lives, languages, cultures, and communities. Because of this, informed consent and community ownership should not be treated as optional. People should have a meaningful role in deciding how their knowledge is collected, used, and represented.
I was also interested in her examples of Inkuba and Masakhane because they show that AI development does not have to copy the priorities of the largest technology companies. Inkuba challenges the assumption that bigger models are always better, while Masakhane demonstrates how research can value many different forms of contribution. These examples make Sambuli’s argument practical rather than purely idealistic.
However, I think the hardest part will be convincing powerful companies and governments to accept limits on their control. Ubuntu may offer a better moral framework, but those with the most resources still have strong incentives to prioritize speed, ownership, and profit. For Sambuli’s vision to succeed, communities must receive real decision-making power rather than symbolic representation. Her speech ultimately shows that the future of AI should not be judged only by what the technology can do, but by whether it strengthens or damages the relationships that make society possible.
LikeLike