Day 2 – Introduction to Large Language Models Lesson

Lesson Overview

Large language models are at the center of generative AI. This lesson explains what they are, how they are trained, and how that training enables them to generate human-like text. You will see why these models can write emails, translate languages, help with creative work, summarize long documents, classify text, and even write computer code. Because training requires vast amounts of data and enormous computing resources, only large companies operate the most capable models. This context helps you understand both the power and limits of the tools you will use in the rest of the boot camp.

You will also learn how LLMs actually produce outputs. They do not “understand” in a human sense. They predict the next word in a sequence based on the patterns they learned from billions or even trillions of words. This explains a key limitation called hallucination, where a model produces confident answers that are not correct. Knowing this upfront helps you use models more effectively and review results with the right level of care.

This lesson is useful whether you want faster emails, clearer writing, help thinking through ideas, quick translations, or starter code. It sets up upcoming hands-on work with ChatGPT and shows where more advanced capabilities like image and PDF analysis come in later.

Who This Is For

If you plan to use AI to write, plan, translate, analyze, or code, this lesson gives you the foundation to work smarter.

• Professionals who draft or reply to many emails and want faster, clearer messages
• Marketers and small business owners who need ideas and structured plans
• Educators and students who need summaries of long readings
• Customer support or operations teams who want consistent, on-tone replies
• Creators and product thinkers who want to brainstorm through conversation
• Developers and non-developers who need code snippets or quick prototypes
  

Where This Fits in a Workflow

Use what you learn here at the start of any task that involves text. LLMs are ideal for a quick first draft, rewriting for tone, or producing a summary before you dig deeper. They are also helpful for cross-language communication and early-stage coding work.

For example, if you receive a long email, you can have a model summarize it, then ask for a short, professional reply. With Google Gemini connected to Gmail, it can even read the thread and suggest a response with more context. If you have a lengthy document, paste the text and ask for a one paragraph summary before you decide what to read closely. When you are stuck on a marketing plan or product idea, use an LLM to brainstorm options and refine through back and forth prompts.

Technical & Workflow Benefits

Before LLMs, you might write and rewrite emails from scratch, translate text manually or with basic tools, or spend hours condensing long documents. With an LLM, you can generate a polished first draft, rewrite in a click for friendly or professional tone, and translate instantly. You still make final edits, but the heavy lifting is much faster.

This approach shines in two scenarios. First, high volume communication like inbox management, where drafting and proofreading eat time. A model can propose a clear response, and you keep control by editing. Second, early analysis of long content, where summarization and simple tables provide a fast overview. Some models can even interpret images or PDFs and surface key data in tables, charts, and graphs, which you will see in advanced parts of the course.

Free tiers let you try these workflows with some limits. Paid plans, often around 20 dollars per month, unlock higher usage and advanced features. Teams plans can add more usage and, in some tools, more privacy options. Even with paid access, keep an eye on hallucinations and verify important details.

Practice Exercise

Try a short, realistic workflow that covers drafting, rewriting, and summarization.

• Choose one real email you need to send this week. Prompt an LLM with the context and ask it to draft a concise, professional message. Then ask for two rewrites with different tones, such as more friendly and more formal. Pick the best one and make any final edits yourself.
• Find a long text you need to digest, such as a multi page document or an article. Paste the content and ask for a one paragraph summary, plus three bullet key points. If you have follow up questions, ask them in the same chat to refine the summary.
• Optional if you use Gmail: if you are comfortable granting access, use Google Gemini with Gmail to draft a reply to an existing thread. Compare the context-aware suggestion with your manual draft.

Reflection: Which step saved you the most time, and where did you still need to correct or fact check the model’s output?

Course Context Recap

This lesson builds your mental model for how large language models are trained, how they generate text, and why they sometimes hallucinate. It connects the promise of LLMs with practical tasks like email drafting, translation, brainstorming, summarization, and starter code. Next, you will start using ChatGPT directly and practice techniques that transfer to other models. Later in the boot camp you will explore additional tools and advanced capabilities such as vision features that can analyze images and PDFs and produce tables, charts, and graphs. Continue through the course to turn this foundation into everyday, reliable workflows.