Introduction and limitations · Retrieval Augmented Generation (RAG) and Vector Databases for Developers

Generative AI for Beginners, lesson 15.
LLMs lack up-to-date and private data.
Can hallucinate.

Limitation	Effect
Training cutoff	Outdated
No private data	Gaps
Hallucination	Errors

Do this in your own tools, step by step. Ticks are saved on this device.

Note limits.
Follow.
Continue.

Lesson?

15.

Limit?

No private data.

0:09 hi everyone welcome to the generative AI for beginners curriculum I am bethan chumba and I'll be taking you through lesson 15 that is retrieval argumented generation and Vector Data Business let's get started in this lesson we'll cover why exactly Unit D why we even need to care about it talk about the components of frag and we'll show you how you can be able to build a full rag application before we conclude the session in our scenario we'll be using aopi as our then we'll be using data from the AI for beginners curriculum the lesson on neural network and we'll have AAR and a cosmos GP for our database and search index currently our llms have a couple of limitations for example the limitation around knowledge around current events if you go in and ask is Queen Elizabeth the second alive the a model will tell you as from my last update on October 2021 yes she was but I do not have F data the other thing is around your own

1:26 personal data so you it was trained using publicly available data so if I want to ask what is the price of CI from a specific store it will tell me I'm sorry as an AI model I cannot tell you the real time PR of specific stores the other limitation is around the context provided as well as data that responses generated might not