Learn ChatGPT
Section outline
-
ALTITI LEARNING SERIES
From first prompt to practical workflows
Learn to ask better questions, turn ideas into useful work, and check AI-generated results with confidence.
BEGINNER FRIENDLY • 4 LESSONS • ABOUT 75 MINUTES • SELF-PACED
Welcome to Learn ChatGPT
This practical course takes you from your first conversation to writing effective prompts, exploring data, and giving an agent clear instructions. Each lesson combines a short explanation, an example, and a hands-on activity. Official OpenAI learning videos support your practice.
Start Lesson 1 → Work through the lessons in order, at your own pace.
What you will be able to do
- Start a useful conversation and improve an answer with follow-up questions.
- Write prompts with a clear task, context, audience, and output format.
- Summarize a small dataset and verify the important numbers.
- Define an agent task with permissions, checkpoints, and success criteria.
Your learning path
LESSON 01 · 15 MINUTES
Get started with ChatGPT
Create your first conversation and practice a useful follow-up.
LESSON 02 · 20 MINUTES
Write clear prompts and improve answers
Give useful context, set constraints, and refine your result.
LESSON 03 · 20 MINUTES
Analyze files and verify the results
Explore a sample dataset and check its totals and conclusions.
LESSON 04 · 20 MINUTES
Delegate tasks to an agent responsibly
Build a clear task brief, then try the interactive video knowledge check.
Before you begin
Have access to ChatGPT and a place to save your practice notes. No coding experience is required. Use the fictional examples provided in the lessons; features may vary by your plan and available tools.
Your course takeaway: an improved prompt, a verified data summary, and a reusable agent task brief. Finish with the H5P video and three-question knowledge check.
An independent ALTITI learning course using linked official OpenAI resources. This course does not award an OpenAI certification.
-
Your first useful conversation
Goal: Start a conversation, give context, and improve an answer. About 15 minutes.
Start with one clear task. Explain who the answer is for and what a useful result should look like. Read the response, check its accuracy, and ask for a revision.
- Start a new ChatGPT chat.
- Ask: “Explain a learning management system to a new teacher in 150 words. Include one classroom example.”
- Follow up: “Make this simpler and turn it into a five-item checklist.”
- Check the result against your needs.
Watch:
. Observe how a clear goal becomes a deliverable. Features vary by plan and tools.Practice: Create a three-day study plan, then revise it to fit 20 minutes per day.
Self-check: What improved after your follow-up?
Video selected from ChatGPT Learn. This independent ALTITI-LMS course is not an OpenAI certification.
-
Give the task, context, and success criteria
Goal: Turn a vague request into a prompt you can evaluate. About 20 minutes.
A useful prompt states the task, the relevant background, the intended audience, and the desired format. Add constraints that matter, such as length or which source to use. Continue the conversation with specific feedback.
Before: “Write something about training.”
After: “Draft a 120-word invitation to a beginner ChatGPT workshop for university staff. Use a friendly professional tone. Include placeholders for date and room, and finish with a registration call to action. Do not invent event details.”
- Write your own version for a real task.
- Ask ChatGPT for a first draft.
- Identify one weakness and request a targeted revision.
- Compare the two versions using accuracy, relevance, and clarity.
Read: Official ChatGPT prompting guide.
Watch and try:
. If voice is available, speak the same task and refine it aloud; otherwise use text.Self-check: A confident tone is not proof of accuracy. Which details in your draft still need checking?
-
From a small dataset to a checked summary
Goal: Ask a precise data question and verify the answer. About 20 minutes.
When file tools are available, provide a document or spreadsheet that you are allowed to share. Explain the columns, the question, and the output you need. Ask the assistant to identify missing values and assumptions before drawing conclusions.
Practice data: Use these invented figures: Monday 12 registrations; Tuesday 18; Wednesday 15.
Prompt: “Summarize these registrations. Calculate the total and daily average, identify the busiest day, and show the calculation. Use only the supplied numbers.”
Check your answer: The total is 45, the daily average is 15, and Tuesday is busiest with 18. If a result differs, inspect the inputs and calculation before using it.
Extension: Put the figures into a spreadsheet with Day and Registrations columns. If file upload is available, upload it and ask for the same summary. Compare it with your manual check.
Before sharing a result: Check row counts, missing values, units, date ranges, and any claims that go beyond the data. Use invented or anonymized data for this exercise.
Video revisit:
. Connect the idea of a deliverable to a report you could use.Self-check: What evidence would convince you that a summary is correct?
-
Define the goal, tools, boundaries, and checks
Goal: Write a clear instruction for a multi-step task. About 20 minutes.
An agent can use connected tools to take actions and inspect results. Access depends on the available tools and permissions. Some tasks need you to complete a sign-in or approval step. A longer workflow should include checkpoints and a clear response to errors.
Watch:
.Example instruction: “Use the provided fictional registrations CSV. Check required columns and duplicate usernames, prepare a UTF-8 CSV, and show a summary of proposed additions and updates. Stop if validation fails. Do not upload or change accounts during this test.”
- State the intended result and exact input.
- Identify the website or application the agent may use.
- Specify permitted changes and any review point.
- Define success: expected rows, saved file location, and a result report.
- Define failure behavior: stop, report the issue, and preserve completed work.
Practice: Write a five-step instruction for a routine task you do each week. Include one checkpoint and one way to verify the final result.
Knowledge check: A completed upload is not enough evidence of success. Compare the created, updated, skipped, and failed counts with the input.
Course finish: Keep one improved prompt, one verified data summary, and one agent task instruction as your personal starter toolkit.
-
-