Mission

Education is one of the areas for artificial intelligence with the highest potential for positive impact, with the opportunity to offer students tailored learning paths with personalized material for marginal cost.

At the same time, particularly school education is one of the least funded areas of innovation at the moment. Governmental organizations oftentimes struggle to understand the potentials while the private sector is faced with enormous fragmentation, high barriers of entry and substantial regulatory risks.

I made it my mission to help education organizations, both non-profit and pro-profit, to overcome these challenges, providing AI models and tools that enable them to leverage the potential of AI-supported learning in schools with a standardized approach to describe competence building.

Approach

The key challenge of a safe adoption of AI in schools is to ground it in facts about the very nature of building competences. Before anything else, humans and AI need a shared perspective on how we acquire core competences like reading and arithmetic or otherwise we cannot reflect on AI decisions and correct them where necessary.

By providing a map of reasonable learning paths, a knowledge graph of competences provides the foundations of safe AI. We can then combine this graph with specifically trained classification model to locate learning material and learning activities on our map. This way we can describe learning journeys across various interactions and safely reason about them.

Further, by making this map an open standard, learning journeys can be shared across different organizations, like schools, learning management systems, textbook publishers and educational games. Having a common language massively simplifies the challenge of maintaining a mostly complete trail of a student's learning path, the prerequisite for meaningful personalization and recommendations.

Providing an open ontology (the map) in combination with open source models for labeling, comparing and reasoning about content (the AI), we can address this shared challenge for all education organizations (non-profit and pro-profit) without relying on a centralized infrastructure.

Roadmap

Proof of Concept

[x] Foundational Ontology
[x] Zero-Shot classification (Gemini)
[x] Fine tuned classification (Gemini)
[x] Similarity/clustering (Vertex.ai)

Open Source Launch

[x] Open Ontology for Primary School Math
[x] Open training set generator
[x] Open supervised training for multimodal LLMs
[x] Open similarity/clustering model
[ ] Demo application (in progress)
[ ] Open reasoning model

Long Term Plans

[ ] Covering reading and writing (primary school)
[ ] Covering math for secondary school
[ ] Covering other STEM subjects
[ ] ...

Background

I am a completely independent individual, not affiliated with any education organization. This initiative stems from my impressions during innovation consulting in the US education industry which gave me a deep insight on some foundational issues of the industry.

Being based in Europe and having a teacher's education for computer science, I am at the same time aware of the even bigger fragmentation there. Overall, everyone in education is facing massive challenges in their attempts of adopting AI due to a lack of foundational standards and models that I help to provide.

Support

Your support will ensure a continuous development of my projects and help bring them to education organizations everywhere in the world.

I have been investing a substantial amount of development without funding to start this initiative, primarily researching how various ontology models interact with well-known multi modal LLMs and embedding models.

Contributions to this project will finance the launch of a minimal viable open source solution package, ready for use in the area of primary school math with plans to branch out from there.

Every sponsor will receive a personal shout-out on LinkedIn (unless you want to stay anonymous) and large institutional sponsors can make their support visible with banners in the README of my projects.

Thank you so much for your support!