Why does the talk argue that data is not an unlimited resource in AI?

The talk reframes data as representing lives, cultures, and communities, not just a commodity. It cautions that high-quality data for training AI is drying up, so governance should emphasize meaningful participation and consent. Start time: 156 seconds.

What is data justice in African AI policy and why does it matter?

Data justice ensures that women and rural communities with unique knowledge are represented and visible in data systems, guiding inclusive solutions and local ownership. Start time: 206 seconds.

What is Inkuba and what makes it notable?

Inkuba is a lightweight African language model developed to serve low-resource languages, with strong sentiment analysis across multiple languages and efficiency benefits. Start time: 261 seconds.

What is Leapa AI doing for African languages?

Leapa AI develops lightweight language models to support African languages without needing vast resources, illustrating a practical approach to low-resource NLP. Start time: 249 seconds.

How many languages did Masahane publish translations for, and what does that imply?

Masahane has published translation results for over 38 African languages, demonstrating an inclusive, collaborative approach to NLP research across many communities. Start time: 329 seconds.

What is the main takeaway about elephants and grass in this talk?

The metaphor suggests that while elephants (powerful actors) roam, the grass (people and ecosystems) must be protected. Relationality means power should serve a healthy environment for everyone. Start time: 392 seconds.

What Ancestral Intelligence Can Teach Us About AI | Nanjira Sambuli | TED

TEDx Talks

People & Blogs3 min read8 min video

Mar 4, 2026|13,245 views|329|34

TEDTalk TEDTalks TED Talk TED Talks TED

Save to Pod

Key Moments

TL;DR

Ubuntu guides AI: power with responsibility, data justice, and grass-first ethics.

Key Insights

Reframes AI power: The elephants (nations, corporations) should not trample the grass (people and ecosystems); instead, power should enable coexistence and healthy ecosystems.

Ubuntu/Utu as a governance compass: AI development must center humanity, reciprocity with nature, and the spiritual or cosmic dimensions of community.

Data justice as core design: Data represents lives and cultures; governance should ensure informed consent, participation, self-determination, and community ownership.

Low-resource innovation matters: Africa demonstrates that high impact AI can come from lightweight models and community-driven approaches, not just massive data and compute.

Collective authorship and local collaboration: Initiatives like Masakhane redefine authorship to include all contributors, enabling translation across many languages.

Indigenous wisdom powering AI futures: Ancestral intelligence (Ubuntk) informs AI that serves the common good and respects ecological balance.

THE ELEPHANTS AND THE GRASS: REFRAMING AI POWER

The talk opens with a vivid metaphor: elephants representing great powers—nation-states, corporations, or oligarchies—who maneuver over scarce resources, ideas, and innovations, while the grass stands for people, places, and ecologies that can be exploited or neglected. Rather than fixating on the elephants’ might, the speaker urges us to focus on the grass and its ecosystems. This shift reframes AI as a relational enterprise where power should be exercised in ways that sustain and enrich the broader environment rather than extract at the expense of communities. By highlighting this ecosystem approach, the talk invites a reimagining of humanity’s trajectory in the intelligent age—one where power serves shared flourishing and where the grass is valued as a living, interdependent domain that deserves protection and respectful engagement.

UBUNTU/UTU: A HUMAN-CENTERED ETHIC FOR TECHNOLOGY

Ubuntu (and its Eastern African counterpart Utu) offers a philosophy that humans are defined by their relationships with others and with the natural world. The speaker extends this to AI design, advocating for technologies that benefit all of humanity and our ecology. Ubuntu calls for data governance built on meaningful participation, informed consent, self-determination, and community ownership of data sets—especially those containing language, nature-based knowledge, and indigenous wisdom. This ethical framework is presented as a counterweight to extractive tech cultures, positioning AI as a tool for collective well-being rather than a means to concentrate power.

DATA JUSTICE: TREATING DATA AS LIVES, CULTURES, AND COMMUNITIES

Challenging the prevailing proverb that data is the new oil, the speaker argues that data should be treated as living legacies with cultural and community significance. In African contexts, data governance should ensure that diverse communities—such as rural women with intimate knowledge of agriculture, medicine, and environmental stewardship—are represented and visible in AI systems. This shift leads to data justice: ensuring participation, consent, and ownership, and recognizing that high-quality data is not merely abundant but ethically sourced and inclusive. By reframing data as lives rather than resources, AI becomes more trustworthy and socially beneficial.

LOW-RESOURCE LANGUAGE AI: INKUBA, LEAPA, AND MASAKHANE

The talk spotlights pragmatic African responses to language resource constraints. Initiatives like Inkuba (a small language model inspired by a dung beetle’s strength) and Leapa AI demonstrate that meaningful NLP work can occur with modest parameters and targeted efficiency. Inkuba, trained on 0.4 billion parameters, achieves strong sentiment analysis and cross-language consistency, challenging the assumption that only large models can perform well. Masakhane, a pan-African collaboration across 30 countries, advances natural language processing and translation, including a non-traditional authorship model that recognizes all types of contribution. These examples illustrate that language technology can be locally relevant, resource-efficient, and democratically produced.

A COLLECTIVE FUTURE: CO-CREATION, AUTHORITY, AND ANCESTRAL INTELLIGENCE

Ubuntk—the idea of AI powered by ancestral intelligence—embodies a co-creative model in which indigenous wisdom informs modern AI practice. The African policy and community-building efforts aim to democratize AI development, granting agency to communities and ensuring that technology serves social and ecological well-being. The talk also underscores a broader ecological ethic: even the elephants can become ecosystem engineers when they align their power with the grass’s vitality, creating room for biodiversity and sustainable growth. The closing message is a hopeful invitation to reimagine humanity together—honoring ancestral knowledge, resisting extraction, and building AI futures that benefit everyone.

Mentioned in This Episode

●Tools & Products

Selected quantitative data about Ubuntu-inspired AI initiatives

Data extracted from this episode

Item	Value	Context
Inkuba parameters	0.4 billion	Language model size
Inkuba inspiration	250x body weight	Dung beetle reference
Masahane languages translated	38	Publication translations
Masahane partner countries	30	Collaborative scope

Common Questions

Ubuntk is a term adapted from Ubuntu that emphasizes humanity, communal participation, and governance. It frames AI as a technology to benefit all of humanity and ecosystems, not just powerful actors. Start time: 153 seconds.

Topics

Ubuntk Data Justice Low-resource Language Models Masahane Leapa AI Inkuba Dung Beetle Indigenous Wisdom Africa AI Relationality

Mentioned in this video

toolInkuba

A small African language model inspired by the dung beetle; trained on 0.4 billion parameters and effective across multiple languages.

toolMasahane

A collaborative AI community across Africa aimed at strengthening natural language processing and inclusive authorship in research.

toolDong Beetle

The dung beetle referenced as an inspiration for Inkuba's design and efficiency.

toolLeapa AI

A regional initiative that develops lightweight African language models to serve communities with limited resources.

Found this useful? Build your knowledge library

Get AI-powered summaries of any YouTube video, podcast, or article in seconds. Save them to your personal pods and access them anytime.

Try Summify free