Azure AI Series :How ChatGPT Really Works! Introduction to Generative AI & Large Language Models #01

Welcome to Session 01 of the JBSWiki AI Learning Series, a 17-session journey to understand and work with Azure AI.

In this session, we explore how Generative AI works, what Large Language Models (LLMs) are, and how tools like ChatGPT generate responses.

Our examples throughout this series use BlueFern Travel, a fictional travel agency that wants to build an AI-powered customer support assistant called Fern.

What will you learn?

In this session, we cover:

  • Artificial Intelligence (AI): Understanding AI, Machine Learning, Deep Learning and Generative AI.
  • Large Language Models (LLMs): What they are and how they generate text.
  • Tokens: How text is broken into smaller pieces that models understand.
  • Next-token prediction: How ChatGPT generates responses one piece at a time.
  • Context window: Why models have limits on how much information they can process.
  • Hallucinations: Why AI sometimes generates convincing but incorrect answers.
  • Microsoft Foundry: An introduction to the model catalog and chat playground.
  • Responsible AI: Understanding security, fairness and human oversight.

We also demonstrate how to count tokens using Python and experiment with a model in the Microsoft Foundry chat playground.

Watch the video

Watch the complete step-by-step walkthrough here:

Download Session 01 Materials

Use the following materials to follow along with the video and complete the exercises.

Session files:

The practice files include:

  • count_tokens.py – Python script to count tokens.
  • use_case_worksheet.md – Exercise to identify suitable and unsuitable Generative AI use cases.

Practice Assignment

After watching the video, try these exercises:

  1. Install Python 3.10 or later.
  2. Install the tiktoken Python package.
  3. Run python count_tokens.py and examine the token counts.
  4. Test three sentences of your own, including one in your mother tongue.
  5. Complete the use-case worksheet by classifying the ten BlueFern Travel scenarios.
  6. Read the Microsoft Learn introduction to Microsoft Foundry.

Note: You do not need an Azure subscription for the initial learning exercises. An Azure environment is required for the Foundry playground demonstration.

What’s Next? – Session 02

In Session 02, we prepare our development environment for Azure AI projects.

We will install and configure:

  • Python
  • Visual Studio Code
  • Azure CLI
  • Git

We will also create the BlueFern project folder structure and make our first REST API call using Python.

Follow the Complete Series

This is Session 01 of 17. Subscribe to JBSWiki to follow the complete Azure AI learning journey.

YouTube: https://www.youtube.com/@jbswiki
Website: https://jbswiki.com

JBSWiki – Helping You Master Technology, One Session at a Time!

Thank You,
Vivek Janakiraman

Disclaimer:
The views expressed on this blog are mine alone and do not reflect the views of my company or anyone else. All postings on this blog are provided “AS IS” with no warranties, and confers no rights.