Name: Offline AI, Open Knowledge: Delivering OER to Schools Without Internet
Start: 2026-10-07T16:20:00-0400
End: 2026-10-07T16:50:00-0400

All sessions are available online except round tables, special activities, and workshops.

Offline AI, Open Knowledge: Delivering OER to Schools Without Internet

Wednesday October 7, 2026 4:20pm - 4:50pm EDT

7 DR5

ID: 31657

Description:Across Africa, and some parts of Asia millions of learners sit in classrooms that are rich in curiosity but poor in connectivity. For these students, the promise of open education , freely accessible, world-class knowledge for anyone, anywhere, remains largely theoretical. The internet is the assumed delivery mechanism for most OER platforms, and where the internet is absent or unreliable, so too is access to open content.AXAM is an offline AI-powered learning platform built to close that gap. Developed through the NSF Innovation Corps (I-Corps) program, AXAM packages high-quality open educational resources, beginning with MIT OpenCourseWare transcripts and expanding to broader OER collections, into a locally deployable system that runs entirely without an internet connection. Students interact with AXAM through a conversational AI interface powered by a lightweight large language model, asking questions, exploring concepts, and receiving contextualized responses drawn from curated OER content. No cloud. No bandwidth. No barriers.This session presents the AXAM model as both a technical case study and a provocation for the open education community. The presenter will walk through the architecture of the system: how OER content is processed, embedded, and stored in a vector database; how a quantized language model runs efficiently on low-cost hardware; and how multilingual retrieval supports learners across English, French, Swahili, and Kinyarwanda and others. Crucially, the session will move beyond the technical to examine what deployment actually looks like in contexts where infrastructure, teacher capacity, and institutional trust are all variables that no algorithm can fully anticipate.The lessons from building and testing AXAM are honest ones. Multilingual performance is uneven; Kinyarwanda retrieval lags significantly behind English, raising important questions about whose languages open AI systems are truly built for. Hardware constraints shape every design decision. Community trust must be earned before any technology is adopted. These are not edge cases; they are the core design conditions for open education in much of the world.What this session ultimately offers is a replicable framework, a set of architectural principles, deployment considerations, and community engagement strategies that any institution, NGO, or open education practitioner can adapt for their own low-connectivity context. The goal is not to present AXAM as a finished solution, but to share what has been learned in the process of building it, and to invite the global open education community into the next phase of that work.Because openness without accessibility is just a promise. And a promise that only reaches those with a stable internet connection is not yet open enough.

Speakers

Emmanuel Olimi Kasigazi

Entrepreneurial Lead, Axam AI

Emmanuel Olimi is a data and LLM engineer, open education advocate, and founder of AXAM, an offline AI-powered learning platform designed to deliver MIT OpenCourseWare and other open education content to students in low-connectivity schools across the world. Born in Uganda and now... Read More →

Wednesday October 7, 2026 4:20pm - 4:50pm EDT
7 DR5 MIT Samberg Conference Center, 50 Memorial Drive, Cambridge MA 02139 USA

Presentation

Speaker Modality In-person
Track Exploring Emergent Technologies and the Future of Openness

Tickets

Success you have shared X tickets.

OEGlobal 2026

Emmanuel Olimi Kasigazi

Get help with the event

Success you have shared X tickets.

OEGlobal 2026

Emmanuel Olimi Kasigazi

Sign up or log in to save this to your schedule, view media, leave feedback and see who's attending!

Get help with the event