How does data determine what artificial intelligence can do?
How is AI trained? Lesson plan
Artificial intelligence is often trained on data we share online. Help students become more critical and responsible users of this technology by gaining a deeper understanding of how AI uses data to learn and create.
- Understand that data is a building block of artificial intelligence.
- Identify what AI can do based on the data it is trained on.
- data – any type of information that can be collected, categorized, and analyzed
- input – the data an AI application uses to learn or to perform a task
- output – the final result or creation of an AI system, based on what it was asked to do and using its existing inputs
What you'll need
Before the lesson
If your students are not familiar with what artificial intelligence is, we suggest teaching the What Is AI? lesson, or showing the What Is AI? video first.
Step by step
- Show Slide 4 and explain that students are looking at an illustration created by generative AI. The image is meant to show how AI learns.
Give students a few minutes to analyze the picture and share their answers to the reflection questions:
- What do you notice?
- How does the picture show the AI learning process?
- Review slides 5–8 and highlight some of the key themes in the illustration:
- The robot in the middle represents artificial intelligence. It's surrounded by people who are talking to it and showing it books. These people are the human "trainers" or computer scientists (Slide 5).
- The "trainers" share data to help the AI learn. This data can be things like text from books, websites, or even human speech (Slide 6).
- Data is any type of information that can be collected, categorized, and analyzed (Slide 7).
- We can use the data the AI is trained on, also known as the inputs, to infer or make an educated guess about the types of things the AI might be able to do, which are the outputs (Slide 8).
Note: Students might notice that all the AI trainers in the illustration are men. You can explain that this is a common bias that generative AI exhibits.
- Define the words input (Slide 9) and output (Slide 10).
- Say: Now that we know about how AI is trained to create a certain output, let's look at some real-life examples (Slide 11).
As you go through each example, allow students time to pair-share and make an educated guess before revealing the answer on the next slide (Slides 12–15).
- Ask students to pair up to go through two additional scenarios. This time, they are given the output and need to think about the necessary inputs.
- Scenario: YouTube video recommendations (Slides 16–17).
- Scenario: Targeted advertisement (Slides 18–19).
- Say: As these last two examples show, what we do online and the data we share is also a part of the input, or data, that AI might use. Thinking critically about the data (or input) behind AI can help us be more informed users and understand what AI's limitations might be (Slide 20).
Have your students keep a journal of the types of data that might be generated by their actions online. Ask them to write down the types of tasks AI could be trained to do based on this data.