TechnicalAI glossary
Retrieval-Augmented Generation
"Retrieval-Augmented Generation": before answering, the AI goes and fetches relevant information from a document base, then writes with it. Fewer hallucinations, with sources to back it up.
In other words
An expert who opens the right archive drawer before answering you, rather than quoting from memory.
Alyriaz articles that mention it
- Why doesn't the same AI respond differently depending on the app?
The same model produces very different responses depending on the application using it. We explain the five factors behind this gap, and why it's useful to know.
- General-purpose or specialised model: which one should you choose for your job?
Faced with the vertical models emerging in law, healthcare or finance, we explain when a specialised model genuinely adds value and when a well-equipped generalist is enough.
- What is homomorphic encryption? Computing on data without ever reading it
Homomorphic encryption makes it possible to perform calculations on encrypted data without ever accessing it in plaintext. We explain the principle, its current limitations, and what it could change for AI and privacy.
- RAG or fine-tuning: which one should you choose to connect an AI to your data?
To adapt an AI to your data, two approaches are at odds: RAG and fine-tuning. We explain the difference, when to use one or the other, and why most projects pick the wrong method.
- What is an "embedding"? The technique that lets a machine understand that "doctor" and "physician" mean the same thing.
An embedding transforms text into a list of numbers that encodes its meaning. We explain this key concept, how it enables semantic similarity search, and why it is the invisible engine behind most enterprise document assistants.
- RAG: Why your AI now knows your secrets
Conventional AI cannot read confidential documents, because they are not part of Wikipedia. The 'RAG' technique gets around this problem.