Artificial Intelligence (AI)

Starting from a question about transitions asked by the user, WeLearn suggests an answer based on relevant and qualified resources (articles, videos, podcasts, people, projects…), enabling the user to access reliable information, acquire knowledge and identify experts.

To achieve this, WeLearn relies on a bank of resources, either open or made available by partner organisations.

Its architecture, based on the RAG (Retrieval-Augmented Generation) model, uses several state-of-the-art technical building blocks:

• LLMs (Large Language Models)  for high-dimensional text embedding and man-machine interaction
• Vector databases for information storage
• ANN (Approximate Nearest Neighbours) search for recommendations

The Learning Planet Institute is currently working with partner organizations to adapt WeLearn to their specific resources for higher education audiences.