Share it

‍

Libraries Without Borders (LWB) is a non profit that facilitates access to information, education, culture and information for people who are vulnerable or affected by humanitarian crises. Present in around thirty countries BSF collaborates with local actors to reach around 500,000 people each year, mainly young people.

Innovation is at the heart of LWB's expertise, it is even its first pillar with the creation of content adapted to each context and the training of mediators.

Its flagship tool, the Ideas Box, designed in 2024 by Philippe Starck, is a mobile media library that can be deployed in twenty minutes on any terrain with books of course but also games, creative equipment, tablets and computers allowing access to digital content even without an Internet connection, thanks to the Ideas Cube. It is a small server designed by BSF to provide access to numerous offline resources.

LWB's social enterprise, Kajou, has also invented a micro SD card that fits into mobile phones and allows access to training content without an internet connection or data consumption.
‍

Using AI in the service of its mandate...

In the same logic of using innovation as a lever to reach the populations who are most distant from knowledge due to economic, social, technical or symbolic aspacts, LWB has just launched its IDEAS AI initiative.

In this context, LWB advocates and acts for using AI at the service of cultural and linguistic diversity, technologies accessible in open science and in frugal technical contexts as well as for the massive training of all citizens to make this technological revolution a revolution in access to knowledge and education in the world.

In concrete terms, LWB is already using AI in projects in three areas to make a difference:

Education:

In countries where the language of instruction is French such as Senegal, an AI-based application is being developed to help teachers master French and thus improve the quality of teaching and therefore student success by providing adapted linguistic resources.

Health:

Health systems in many African countries, such as Côte d'Ivoire and Senegal, need simple and practical tools to improve the training of health professionals and the quality of care.

LWB and Kajou are therefore working on an AI-based virtual assistant in order to support community health workers in terms of continuing education and medical practice. The model is currently being trained and LWB is working on the first user stories for the application.

The law:

In regions where sexual violence in conflicts is a real scourge such as the DRC or Nigeria, LWB is developing an intelligent interface based on a vast legal database, thus helping professionals in the field to support survivors in their efforts and reconstruction.

‍

... And strengthen its internal capacities

As mentioned with the Solinum case in a previous newsletter, LWB also uses AI to boost its internal operations and maximize its impact:

1. For data management: using AI to analyze sourced data.
2. To create visual content: with tools like DALL-E and MidJourney producing images for people who can't read.
3. And finally to translate and generate scripts for educational videos: thanks to AI, the association is accelerating the production of educational content adapted to its various audiences.

LWB sees AI as a chance to transform the use of technology in developing countries, where internet access is often limited. Their ambition: to create AI models that are accessible offline and directly integrated into their tools to maximize their impact.

‍

Collaborations and challenges: towards more inclusive AI models

LWB is facing significant challenges such as the lack of resources and local infrastructures not being powerful enough to run the largest AI models as well as the lack of support for African dialects by current language models which are mostly trained with European languages - which generates many errors. If you are interested, we recommend that you read our article on how LLMs work.

These limitations represent a major obstacle for LWB, whose mandate is mainly focused on populations outside Europe. The objective is therefore to develop language models in local languages so that AI is at the service of linguistic and cultural diversity.

LWB is working hand in hand with two researchers from Pleias, a young start-up specialized in the development of frugal language models, i.e. less resource-intensive, in order to be able to deploy them on local devices. The start-up is distinguished by its ethical and inclusive approach: it uses fully open data, guaranteeing total transparency and essential auditability in the context of education and health. Pleais' teams are also experts in creating efficient models for languages with few resources, such as Wolof and Pulaar, and know how to deploy their models on restricted and local infrastructures.

This approach raises awareness of the importance of ethics in AI and aims to not leave the field open only to the biggest players.

Want to know more about LWB's Ideas AI initiative?

https://www.bibliosansfrontieres.org/ideas-ai/

And if you too use AI in the service of the general interest, contact us to share your experience!

‍