Generative AI Fundamentals

Understand how generative AI creates text and images, what it does well and badly, and how to check it and use it honestly.

Beginner · Generative AI · about 8 hours · 4 modules · certificate on completion

Taught by Chloé Tremblàŷ, Zinhlée Khumàlo, AI Learning Companions.

About this course

Generative AI can draft an email, explain a hard idea, write code or paint a picture from a sentence, and it can also state things that are confidently wrong. This course explains, in plain language, how it works and how to use it well. You'll learn what generative AI is, how a large language model predicts text one token at a time, how that predictor becomes a helpful chat assistant, and how image generators build pictures out of random noise. Then you'll test what it does well and where it fails, learn why it sometimes makes things up, and practise checking its answers against reliable sources.

The course ends with the habits that make AI use responsible and honest: keeping private information out, watching for bias, saying when you've used AI, and staying responsible for anything you share. Every module has a hands-on lab in a chat assistant such as Claude, and the course finishes with a project: an explainer you write with AI help, check claim by claim and disclose honestly.

What you'll learn

  • Explain what generative AI is and how it differs from other kinds of AI
  • Describe how language models produce text one token at a time, and how image generators turn noise into pictures
  • Judge which tasks generative AI handles well and where it's likely to fail
  • Spot likely hallucinations and check AI answers against reliable sources
  • Use AI chat assistants responsibly: protect private information, say when you've used AI and stay accountable for the result

Skills you'll gain

Generative AI, Large language models, AI chat assistants, Prompting basics, Fact-checking, Spotting hallucinations, Responsible AI use, Image generation concepts

Tools

A chat assistant such as Claude

Before you start

None: no maths or programming needed. You need a computer or phone with a web browser and an account with a chat assistant such as Claude (claude.ai); a free account is enough for the labs.

Modules

  1. Meet generative AI

    What generative AI is, how such models learn from data, and how to hold a useful first conversation with a chat assistant.

    • What generative AI is, and what it isn't (10 min)
    • How generative models learn (12 min)
    • Talking to a chat assistant (10 min)
    • Hands-on lab: Explore a chat assistant

  2. How generative models create text and images

    Look inside the machine: tokens and next-token prediction, how a text predictor becomes a helpful assistant, and how image generators turn random noise into a picture.

    • Tokens and next-token prediction (12 min)
    • From text predictor to helpful assistant (12 min)
    • How image generators work, in outline (12 min)
    • Hands-on lab: Inside the model: prediction, context and images

  3. What generative AI does well, and where it fails

    Find the tasks where generative AI is genuinely useful, the ones where it's unreliable, and why it sometimes makes things up with complete confidence.

    • What generative AI does well (10 min)
    • Where generative AI struggles (12 min)
    • Why models make things up (12 min)
    • Hands-on lab: Test the strengths and limits

  4. Checking facts and using AI responsibly

    Build the habits that make AI use trustworthy: checking claims against good sources, keeping private information out, watching for bias and being open about where AI helped.

    • Checking AI answers (12 min)
    • Privacy, security and safe use (10 min)
    • Honest and fair use (12 min)
    • Hands-on lab: Fact-check a briefing and disclose AI use

Project: An AI-assisted explainer, checked and disclosed

Choose a topic you know or can research well and a reader who needs it explained: for example, how to save energy at home for a residents' newsletter, or how volcanoes form for a class of 11-year-olds. Use a chat assistant to help you plan and draft a one-page explainer of 400 to 600 words, keeping a log of the prompts you used and what each produced, and improve the draft with follow-ups. Then check every specific claim against reliable, independent sources, fix or remove what's wrong, and edit the text into your own voice. Finish with an honest AI-use note to publish alongside it, and a short reflection on what the assistant did well, where it went wrong and how you kept private information out.

Part of

The products this course names are trademarks of their owners, who haven't made or endorsed it.

Start the course free