Difficulty: Explorer (Level 1)
Artificial intelligence (AI) is trained on information from the internet and other sources. If its sources have a bias, the AI chatbot might also show that bias.
For example, if you have only ever seen and eaten chocolate ice cream, you might think it is strange when someone offers you strawberry ice cream. You might prefer the chocolate flavour because that is what you have tried before. But that doesn’t mean there is something wrong with the strawberry ice cream.
In the same way, AI chatbots might have been trained to see things in one way. But that doesn’t always mean it is the truth.
Use the activities below to explore how you can think critically about what AI tools tell you.
What’s so scary?
-
Draw four monsters. Use different colours and features for each one. Try to make them as scary as possible. Then, give each monster a name.
-
Bring your monsters to different people, and ask them to pick the monster that they think is the scariest. Keep count of their votes in a separate notebook.
-
After asking at least 10 people, which monster is the scariest? Did everyone pick the same monster?
-
Reflect on how this shows us AI’s limits. What is “scary” can depend on each person’s fears. If AI is trained to find one monster to be scary, that doesn’t mean everyone will be afraid of it. Why should we be careful with information from AI tools?
What’s so tasty?
These days, some people rely on AI for recipes and food recommendations. But just like we learned in the previous activity about what makes a “scary” monster, food and taste are also special to each person, family and community.
-
Ask three people in your neighbourhood about their favourite place to eat nearby. Why is this their top spot?
-
Next, ask an AI chatbot to suggest three places to eat in your area.
-
What is special about the suggestions from real people? How are they different from AI’s recommendations?
