AI Is Learning. But About Africa? | Francis Odeyemi | TEDxOsogbo

AI Is Learning. But About Africa? | Francis Odeyemi | TEDxOsogbo

by Francis Odeyemi · View the channel on PSI

Humanities, Social Sciences & Thought AI & Society UBDigital and Information technology: general topicsUBJDigital and information technologies: social and ethical aspects
🎙 Francis Odeyemi 👥 44.7M 📅 October 19, 2026 ⏱ 11 min 👁 1K 📄 expert opinion 🧭 2026-10-20
Available in: English (current) Français

Keywords

dataAI trainingAfricarepresentationcontent creation

Summary

In this TEDx talk, Francis Odeyemi, founder of Ninthgrid, discusses the severe underrepresentation of African data in the datasets that train large language models (LLMs). He illustrates this with a personal example: asking AI to generate an image of Okefia in Oshogbo yields an incorrect result because AI lacks African context. He cites that only 5% of global LLM training data comes from Africa, and less than 0.5% from Nigeria. He argues that data is power and whoever controls data shapes perception in the age of AI. He traces the shift from traditional blogs to chatbots as primary sources of information, and warns that if Africa is not in the data, it is not in the future. He shares findings from a user research with 40 of his platform’s users, showing that many lack prompting skills and AI models are not trained on African conversational nuances. He urges individuals to contribute by creating authentic African content, participating in data labeling and annotation, and ensuring data is legally sourced. He emphasizes that if Africans do not teach AI about Africa, others will do so in ways that misrepresent the continent. The talk concludes with a powerful statement: if Africa is not in the dataset, it is not in the conversation, and not in the future.

214 words

Critical Evaluation

Value of the Information & Strength of the Argument

The value of the talk lies in its clear articulation of a critical issue: the data bias in AI and its implications for African identity and representation. The speaker uses concrete examples and personal experience to illustrate the problem, making it relatable. His argument is logically structured: he defines AI’s components (infrastructure, algorithm, data), highlights the data gap, and provides actionable solutions. He also touches on the historical shift in information consumption and the role of content creators. However, the argumentation is more persuasive than rigorous, relying on anecdotal evidence and broad statements rather than citing specific research or data sources. The call to action is clear and empowering, but the scientific foundation is thin.

Scientific Rigor, Source Quality, Title Accuracy

The talk is a TEDx presentation, which typically offers a platform for personal insights rather than peer-reviewed research. The speaker mentions statistics about African data representation but does not provide citations for them. The only external source mentioned is the TEDx website link in the description, which is not directly related to the claims. The title accurately foreshadows the content, and the talk stays on topic. There is no evidence of misleading information, but the lack of verifiable sources weakens the overall rigor. The speaker’s expertise as a product designer and founder adds credibility to his practical observations but not to the statistical claims.

234 words

Title / Content Match

The title accurately reflects the core theme: whether AI's learning includes African perspectives, and the speaker addresses this with concrete examples and calls to action.

Quality & Reliability

6/10

The talk presents a compelling argument about underrepresentation of African data in AI training, supported by some statistics (e.g., 5% of global LLM data from Africa, less than 0.5% from Nigeria) but without citing specific studies or datasets. The claims are largely anecdotal and based on personal experience, not rigorous research.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

  • TEDx — Organizational platform for the talk, referenced in the description.

Contribution & Novelties

The talk brings attention to a significant yet underexplored issue: the near absence of African data in AI training sets. It frames data as a form of power and calls for collective action from Africans to shape AI’s understanding of their culture. The speaker provides practical steps, such as creating local content and participating in data annotation, which is a novel angle for a general audience.

Pour aller plus loin :

  • Data annotation — Relevant to the speaker’s suggestion of becoming data labelers.
  • Algorithmic bias — Directly relates to the misrepresentation of Africa in AI outputs.
  • Artificial intelligence and data rights — Discusses legal aspects of data ownership mentioned in the talk.
  • African languages on the internet — Context for the lack of African language data in AI systems.

129 words

Radar Profile

The profile shows moderate scores across all dimensions, with slightly higher quantity of information and lower technical depth, reflecting a persuasive talk that offers many ideas but lacks deep scientific rigor.

Reliability 6/10