Kolibri

Over half of the world is now on the Internet, but for those who are still offline, the lack of infrastructure constitutes a huge barrier for accessing learning. Kolibri is Learning Equality’s adaptable set of open solutions designed for offline-first teaching and learning to bring quality learning experiences to disconnected learners, and to provide tools for teachers to best support their students. During the pandemic, the need to support learners not just in schools, but also at home became clear, and we are now committed to enhancing Kolibri to support at-home, hybrid, and in person learning. This will allow learners with no Internet access to learn at home, while still benefiting from teacher support through remote facilitation. If this was scaled globally, then hundreds of millions of disconnected learners would be able to continue learning effectively during the ongoing pandemic, and would help to mitigate and close the homework gap. 

Due to the current pandemic, schools in many parts of the world have remained closed for more than a year, further exacerbating the learning crisis. When schools eventually start to reopen, it is expected that most schools will need to adopt a hybrid model where learning happens both at home and in classrooms. As a result, the relevant infrastructure, including both Internet connectivity and devices, are not always available to access digital learning materials that support the continuity of learning in this hybrid model. There is now an urgent need for openly available learning materials that are aligned with local curriculum and low-cost solutions for disseminating these materials to students for use at school and at home. In addition, teachers are now required to facilitate learning in a learning model for which they were never trained and for which they have little experience. Providing appropriate support for teachers, means giving them access to training, supportive technology, and effective processes for supporting learners remotely, and at uncertain time intervals.

Learning Equality develops, maintains, and supports Kolibri, an end-to-end suite of open-source tools, content, and do-it-yourself support materials, designed to support learning for the half of the world without Internet access. 

This summer we’re releasing new functionality to support offline-first access to learning with remote teacher facilitation. This will enhance our two-way data synchronization to support remote facilitation for learning. Through Kolibri’s completely offline access and distribution model, we are enabling educators to monitor and offer support to students -- both through Kolibri and through side channels, such as SMS or WhatsApp.

How does this work? Learners are given a device (e.g., tablet) pre-loaded with relevant content aligned to the national curriculum - including videos, simulations, and exercises assigned by their teacher. When they learn at home, their data will be logged on the device. When they are able to come into school, this data will synchronize with the classroom server, allowing their teacher to use Kolibri's existing differentiated instruction tools to monitor their progress through intuitive dashboards, provide feedback, and intervene to help their learning. Additional lessons can then be assigned to continue learning at home, allowing these students to focus on the most effective learning pathway for them.

Kolibri is designed to benefit teachers and learners across grades K-12 with a particular focus on lower secondary learners in marginalized communities around the world. The lack of access to quality learning opportunities for these learners have created widening equity gaps in learning outcomes. The situation has been further exacerbated due to limited access to digital technology and lack of limited support to educators as they transition to hybrid learning.  

We work closely to co-design the product with a core network of collaborators, including national NGOs, UN agencies, government, and corporate partners, while also taking a needs-based approach to its development through insights gathered from Kolibri’s global user community. For example, co-design sessions with refugee and host community teachers in Kakuma Refugee Camp informed Kolibri content discoverability, user interface design for the learner, resource selection, and modalities for accessing Kolibri across the camp. Learning processes were shared across countries: refugee coaches from Kenya trained teachers in Uganda, and Ugandan Ministry teacher trainers supported professional development and lesson sharing for refugee coaches in Tanzania. 

We've always focused on boosting learning outcomes in some of the world's most challenging contexts, supporting disconnected learning in refugee camps, rural schools, orphanages, and out-of-school programs. Since we take a needs-driven approach to product design, the combination of co-design and working through implementing organizations enables learners to acquire the relevant skills to support their livelihood including preparation for exams, acquisition of relevant certifications, and skills for employability.

Low-connectivity does not mean no technology. Our focus is on pedagogies to blend technology in learning: encouraging activity outside of the platform through project-based learning, sharing lesson plans for teachers, supplemental learning resources aligned for self-study, and interactive activities to simulate what may not be possible in low-resource contexts. Kolibri is designed to support teachers with limited capacity and training, users with limited digital literacy skills, and learning in large class sizes where there may be differing learning abilities within a single class. Overall, this adaptable and flexible approach allows users to contextualize based on needs and address equity constraints. 

The Kolibri Product Ecosystem is already scaling, with a content library containing materials from 40+ content providers, global use of the learning platform, a toolkit to support learning in formal and nonformal contexts, and a robust curricular alignment tool that aligns digital resources to national curricular standards. However, the growth development stage applies because the database replication mechanisms we’re developing to support individual syncing of learner data across different installations of Kolibri will be incorporated in our next major release this summer. It requires a robust set of training and guidance materials to support its use, building on the distance learning materials developed during COVID. This new functionality will be piloted across multiple communities in Honduras. Once tested and adapted, it will be incorporated in the next major release of Kolibri; from there, we’ll continue to iterate as Kolibri scales, given the expanded ability to support learners on the go.

Kolibri is innovative because it works within and responds to the infrastructural realities in low-resource environments, while focusing on the learning needs in these contexts. It is an adaptable solution created for and responsive to the diverse learning needs in these environments. It can be leveraged in contexts with limited electricity, and where the Internet is costly and/or not prevalent. 

Kolibri is low maintenance: while it is not low-tech, it is a robust learning platform extensively tested in low-connectivity contexts that does not require significant digital literacy skills to use; educators and learners with limited exposure to technology are able to get started with Kolibri swiftly. 

Kolibri responds to the learning needs where it is being used: it leverages hardware that already exists, from networking infrastructure to existing computer labs, and also runs on a variety of low-cost devices.

Its feature set and flexible hardware model responds directly to limited teacher capacity and training, limited digital literacy, large class sizes, and differing learning abilities among a group of learners. Lastly, it is open-source, which is critical to ensure that learning needs, particularly for the most marginalized learners, are met: it is free and open, it can be used and adapted for one’s own needs, it can benefit from open contributors to build and improve the product, and new functionality is driven and prioritized by the feedback and needs from Kolibri's global users.

The peer-to-peer syncing functionality is a cornerstone of Kolibri’s ability to respond to the needs in disconnected contexts.

Given the offline nature of our solutions, we only ever know about a subset of the usage (based on downloads, update pingbacks, and direct feedback and reports from users and partners). Our first platform, KA Lite, launched in 2012, had reached over 6 million users by 2018. We’re estimating that Kolibri has reached well over 3 million users since its launch in December 2017, but this is a conserative estimate as we continue to develop methodologies for estimating the portion of the user base we are able to measure directly through known collaborations and telemetry data.

With the public launch of our Android application, we expect to open up new channels for rapid growth, and by the end of next year would like to have reached 10 million learners in total, and through rising governmental and public adoption, reach 50 million users over the next 5 years.

The new solution outlined here, with syncing of individual learner data, not only increases the usefulness of our tools for scaled use within institutional contexts (thereby boosting adoption and usage), but through the enhanced data synchronization abilities also improves our ability to collect data, estimate our full reach, and understand user feedback and needs to iterate and adapt.

Learning Equality monitors the success of Kolibri primarily by observing increases in access to quality learning resources and tools, as well as improvements in educator and learner confidence, development of skills, and boosts to learning outcomes.

We also monitor availability of relevant materials as an indicator of access to quality learning opportunities. This will be measured by the increased availability of learning materials that are available in a language learners understand and is consistent with the education policy specific to the beneficiary group, as well as understanding if there are any barriers to accessing these materials (both physically and via the Kolibri Learning Platform itself). Increased availability of aligned materials helps to further reduce equity gaps for marginalized learners. Sourcing aligned content and adapting existing documentation to enable effective implementation program rollouts can be a promising model in contexts around the world. 

Lastly, we collect quantitative and qualitative data on teacher and student engagement to drive future designs and predictive models to support targeted recommendations on Kolibri. Measurement includes periodic surveys and/or interviews depending on access with key questions pertaining to the ease with which the technology is used, the usability of the Kolibri Learning Platform to support learners, and how the technology is supporting teachers to support learners. 

20 full-time employees, 2 part-time employees, and 7 contractors

Our team of more than 20, headquartered in San Diego with staff globally, has been deeply embedded in fighting for equity in learning for 7+ years. Our founding team includes PhDs in Cognitive Science specifically focused on Learning Sciences. They cultivated the team from the ground up, bringing in individuals with expertise in teacher training, international educational development, engineering, curricula, and accessibility. Most notably, for a nonprofit focused on education through the lens of technology, our Product Lead is both a software developer and a former teacher/trainer. While our team is not representative of those who we serve, we leverage our experience in these areas to develop an adaptable set of solutions that can be locally contextualized. We work closely to co-design the product with a core network of collaborators and their beneficiaries, including national NGOs, UN agencies, government, corporate partners, and grassroots organizations. We take a human-centered design approach to the development of the platform, aggregating and generalizing over insights gathered from Kolibri’s global user community, and then validating assumptions with early testing and interviews with users. We develop our software in the open, on our Github repositories, allowing technical users to follow and test new developments, as well as contribute to bug fixes and new features.

Our team is cognizant that as we are a white-led non-profit organization based in the Global North, we have a particular responsibility to address the structures and institutions that perpetuate inequalities at home and through our work. Our team employs a distributed management structure which enables and encourages leadership across the team, and we’re also aware that even within this structure we could be perpetuating the implicit norms of a white dominant culture. As a learning organization, we continue to listen, introspect, and learn, and our whole team engages in critical team dialogue in discussions related to identity and power. As an organization, we also implemented rigorous processes to eliminate biases in our hiring practices, and we have worked with consultants to help us on our journey towards being an actively anti-racist organization. One of our tangible goals for the year is to ensure that our Board of Directors is on this journey with us as well, and to continue to build a Board that is more broadly representative of the voices and communities of our collaborators and beneficiaries around the world.

We believe that we can only tackle the challenge of global educational inequity through collective action and by working closely with stakeholders, communities, experts, and other innovators. We are committed to designing and building our open-source tools in a collaborative and inclusive way. Through the Solve global community, we hope to leverage the expertise of like-minded innovators who are working to tackle global challenges and continue to improve upon our solution. Given that we’re breaking the mold as to what is possible with education technology, gaining exposure in the media and at conferences would help us further our reach and more clearly communicate our work and impact.

Also, in 2021, we are doubling down our efforts to support monitoring and research with Kolibri, and are hiring an impact and evaluation specialist. For Learning Equality, ensuring that we have strong impact measurement and evaluation is critical but also challenging working in environments with disconnected populations and across varying contexts in terms of hardware, pedagogy, content, language, and implementation. Solve can help Learning Equality with connections to research institutions and expert consultants to influence our research strategy and strengthen our impact measurement practice.

Learning Equality’s core network primarily consists of the community of Kolibri users, and key stakeholders that intersect with the Kolibri Product Ecosystem including content creators, hardware providers, and implementing organizations. Gaining additional insights to help to expand our network of supporters and those who are mission aligned is beneficial to the sustainability of developing open source technology. As mentioned above given our current focus on data, particularly as we’re developing the Kolibri Data Portal (a central, online service for aggregating, reporting, and exploring learner data, to guide and support implementation of projects with multiple installations of the Kolibri Learning Platform), having additional researchers and contributors to this development will help with large-scale implementations of Kolibri in the future.

Through the Solve network, we are primarily interested in collaborating with researchers and experts that can support our evolving understanding of impact for each element of the ecosystem. MIT as an institution would be a fantastic collaborator for us (we have previously connected with the MIT Media Lab with their work on refugees and leaders at MIT focused on open learning, and through MIT OpenCourseWare), but having access to MIT researchers to support how we’re measuring the impact of teachers and educators via our solutions would be immensely helpful at this stage of our work. We’ve been fortunate to connect with Solve members in the past through introductions, events and organic engagement, and hope that this continues, particularly with organizations developing learning resources that have local relevance in a country context where there is demand for Kolibri but digital resources are a constraint. We’re also keen to partner with organizations focused on social-emotional learning, like Amal Alliance (who we're exploring a collaboration with), as this will be core to post-pandemic learning.

The core focus of the Kolibri platform is equitable access - in the U.S., and even more pointedly during the COVID-19 pandemic, the disparities in Internet connectivity, and thus to educational resources was put into stark relief. Kolibri’s offline first focus, providing access to resources regardless of Internet connectivity. The ASA Prize would allow us to build on work that has been started during the COVID-19 pandemic, where Kolibri has been used across the state of Pennsylvania to support remote learning for those disconnected from the Internet. This prize would enable us to integrate contextual hints and guidelines on effective teaching into the platform to make professional development a continuous process, allow us to expand our resource library and bring in additional career guidance resources into the platform. It would also allow us to continue our long term development goal of building the platform to fully support project based learning - with initial updates to allow student submission and sharing of a range of file types and coach tools to provide qualitative feedback on student submissions.

Learning Equality has been committed to supporting refugee and host community learners with skills development since its early days with KA Lite. More directly, Learning Equality and UNHCR have collaborated since 2018 to rapidly expand educational opportunities for refugee and displaced people. Recognizing that the majority of refugees are hosted in places which lack the necessary infrastructure and connectivity to support basic education, and educators often face large class sizes with limited time to identify relevant resources, the partnership sought to find solutions to these barriers using Kolibri. Learning Equality is continuing to work with UNHCR in Kenya, Jordan, Uganda and Tanzania, and expanding in 2021 to Libya with potential to support the expansion of Instant Network Schools programme in the DRC and Mozambique in the future. Refugee resilience is promoted through this programme as refugee and host community educators and learners co-designed the program implementation process, resources to support skills acquisition and improved learning outcomes are organized in a way that makes them easy to find, and the platform and training materials are continuing to be adapted to reflect individual identity in the learning experience.

The Andan Prize would be used to fund project management and in-country staffing costs, training programs, content scoping and modification in addition to efforts to align to national curriculum, and ongoing development of the Kolibri Product Ecosystem (including user research to support the iterative design process).

Learning Equality’s work benefits learners around the world, including women and girls. Through telemetry data, we are able to ascertain that we are slowly moving towards achieving gender parity with Kolibri, and so we are working directly with organizations to achieve that end. Some examples include:

• We have been working with UN Women since December 2017 on UN Women’s SCE program to empower women economically and change attitudes and social norms by making learning more accessible. The programme, being piloted in Cameroon, Jordan, India, Mexico, Chile, and Australia, aims to directly benefit young women from indigenous, refugee, displaced, and low-income groups. Kolibri is used to supplement face-to-face trainings in practical skills and entrepreneurship, and help women get back into formal education. Implementing organizations provide support at independent learning centres where women can access Kolibri.

• Kolibri has been identified as a promising practice in Uganda to support transition from primary to secondary education for girls at scale through its work with UNHCR. They’re distributing tablets with the Kolibri Android App for a pilot of 2,000+ learners to minimize learning loss and prepare for exams during the pandemic. 

• Shoulder to Shoulder in Honduras is working with the IADB, IPA and Learning Equality to conduct a randomized control trial on the use of Kolibri to evaluate school retention 2021. During the initial sample, disaggregation by gender was considered to ensure a representative sampling. 

• Kolibri is organically leveraged by 60 million girls in their Mobile Learning Labs for their grantee organizations.

The vast majority of data driving current science of learning research and educational data mining is collected from online platforms (Fischer et al., 2020). As with most academic research into the human mind, this data is heavily biased towards richer learners in industrialized nations (Henrich et al., 2010) because these learners have higher rates of Internet access .

Given this stark imbalance in where data is sourced, lessons learned for design of platforms, changes to practices in teaching and learning, and machine learning models trained on these datasets will tend to be implicitly biased towards those who have Internet access. Learning pathways, personalized learning, and promising practices for teaching and learning that are based upon these data will therefore also be biased.

With this new functionality for the solution in a distributed model, having this syncing makes it viable to support both access and data collection when devices are offline and away from a central offline server such as in a school. This central server will periodically sync to the online Kolibri Data Portal, a centralized location for aggregation and exploration of Kolibri learner data, when connectivity is available. This will then allow these aggregated data to be used as the basis for model training for content recommendation, to develop models that can then be redistributed to offline instances of Kolibri to guide student learning.

The vast majority of data driving current science of learning research and educational data mining is collected from online platforms (Fischer et al., 2020). As with most academic research into the human mind, this data is heavily biased towards richer learners in industrialized nations (Henrich et al., 2010) because these learners have higher rates of Internet access .

Given this stark imbalance in where data is sourced, lessons learned for design of platforms, changes to practices in teaching and learning, and machine learning models trained on these datasets will tend to be implicitly biased towards those who have Internet access. Learning pathways, personalized learning, and promising practices for teaching and learning that are based upon these data will therefore also be biased.

With this new functionality for the solution in a distributed model, having this syncing makes it viable to support both access and data collection when devices are offline and away from a central offline server such as in a school. This central server will periodically sync to the online Kolibri Data Portal, a centralized location for aggregation and exploration of Kolibri learner data, when connectivity is available. This will then allow these aggregated data to be used as the basis for model training for content recommendation, to develop models that can then be redistributed to offline instances of Kolibri to guide student learning.

Learning Equality’s core principles are focused on supporting student-centered learning in STEM. Through Kolibri, learners can engage with STEM resources in the form of videos, exercises, and interactive games/simulations to build foundational concepts. Learners can access quality math resources in a variety of languages through our curated library of resources which includes videos and exercises from Khan Academy, CK-12, PHeT, PraDigi, and others. Marginalized communities have access to these learning resources to help them overcome barriers to learning and to support future career pathways in STEM.