Blog

How might low- and middle-income countries co-design solutions that align with education priorities and ethics?

EdTech Hub’s AI Observatory is exploring how education systems can create context-driven solutions – which means ensuring AI tools, models, or approaches genuinely meet the diverse needs of a place and the people within it.

This week, in Issue No. 29 of our Waypoint Wednesday series, we spotlight initiatives that are aiming to create context-driven solutions by involving stakeholders in the design process, from coordinating to unlock local context, to homegrown collaborations around SLMs, and efforts to embed rights into AI governance.

Early signals

Unlocking local context for more relevant outputs

Local educational data is often scattered and fragmented. We’re seeing coordinated efforts to unlock the context embedded in local data which could help make AI outputs more relevant and culturally aligned.

  • Africa – Unlocking Data: This initiative seeks to unlock the patchwork of local data that would grow local capacity for analysis. Although not explicitly related to AI, this paves the way to more locally-informed datasets. (Unlocking Data)
  • Latin America – Latam-GPT: Latin America’s GPT is trained on high quality local sources rather than general web data, including university theses, digitised books and congressional transcripts, based on the idea that cultural context is embedded in these regional documents. (Latam-GPT)

Homegrown collaborations around Small Language Models (SLMs)

We’re seeing homegrown collaborations form around Small Language Models (SLMs), closely attuned to local priorities. Unlike LLMs which are often unsustainable for resource-constrained budgets (Wang et al., 2024; Naveed et al., 2024), SLMs can be more efficient and adaptable (Brookings, 2025).

  • Southeast Asia Southeast Asian Languages in One Network (SEA-LION): With an open-source, community-driven ethos, SEA-LION is designed to reflect the linguistic and cultural nuance of under-represented communities and low-resource languages across Southeast Asia. (SEA-LION)
  • Africa – InkubaLM by Lelapa AI: Lelapa AI’s InkubaLM-0.4B is an SLM with applications tailored to low-resource African languages. Designed as a smaller, more efficient model and trained across languages including isiZulu, Yoruba, Swahili, and isiXhosa. (Brookings, 2025

Embedding rights in AI governance

We’re seeing efforts to formally include the rights of often overlooked stakeholders in AI governance, ensuring that those impacted by these tools have a voice. 

  • Latin America and the Caribbean – UNESCO guidelines for indigenous data sovereignty: UNESCO’s work on Indigenous data sovereignty in AI promotes the inclusion of Indigenous communities in AI development and affirms their rights over how their data is governed and used. (UNESCO, 2023)
  • United Kingdom – Child-Centered AI: The Children and AI project at the Alan Turing Institute studies how children understand and experience AI, and brings their perspectives into research, policy, and design. It works with children directly to help ensure future AI systems and governance better reflect their rights, needs and wellbeing. (The Alan Turing Institute, 2024)

Reflections:

  • Contextual design needs to go beyond cultural considerations (such as local languages, food and historical sites) to epistemic considerations (such as oral traditions, different forms of reasoning, what counts as knowledge and whose knowledge is centred).
  • Co-design may place participants at risk when unequal power relations shape what is shared and how it is used. Local knowledge can become an extractive resource when for-profit AI systems capture value. Participants contribute labour and insight but retain little control or ownership over outputs, accountability, or downstream harms.

Swipe for a quick take 👇🏽


We’d love to hear from you! What’s been shaping your thinking on AI? Drop your thoughts (and reading recommendations) in the comments. Explore more from EdTech Hub’s AI Observatory.

EdTech Hub’s AI Observatory is made possible with the support of the UK’s Foreign, Commonwealth and Development Office.

Connect with Us

Get a regular round-up of the latest in clear evidence, better decisions, and more learning in EdTech.

Connect with Us​

Get a regular round-up of the latest in clear evidence, better decisions, and more learning in EdTech.

EdTech Hub is supported by

The findings, interpretations, and conclusions expressed in the content on this site do not necessarily reflect the views of The UK government, Bill & Melinda Gates foundation or the World Bank, the Executive Directors of the World Bank, or the governments they represent.

EDTECH HUB 2026. Creative Commons Attribution 4.0 International License.

to top