AI to the Rescue: Finding Lost Pets

Duration: 45 minutes

Grade Levels: K-5

How would artificial intelligence identify a lost pet? In this version of the classic “Guess Who” game, students examine the process that AI uses to make predictions.
Resources

Lost Pet Cards 

Game Boards

Grade Levels: K-5

Duration: 45 minutes

Concepts/Skills: Artificial intelligence, image recognition, pattern recognition, confidence levels, machine learning

Objectives:

  • Explore the basics of artificial intelligence (AI) and understand how it works.
  • Examine how AI can make predictions when animal identification is needed.
  • Provide information and descriptions to support identification of an image. 
  • Express a confidence level that their guess is accurate.

Background Information

Artificial Intelligence (AI) most often refers to a device or program designed to mimic aspects of human intelligence to complete complex tasks and make decisions. AI is also used for:

  • visual recognition (facial recognition, visual image search). 
  • speech recognition (used in software like Siri and a virtual assistant like Amazon's Alexa). 
  • machine translation (i.e., Google Translate).

Machine learning (ML) is a branch of AI where the goal is to create a program that improves over time or “learns” as it processes more data.

Engineers provide training data (e.g., numbers, photos, text) which the AI uses as the basis for an algorithm: the set of rules for making classifications or predictions. Engineers could then provide more data, rules, or both to test or adjust the algorithm to ensure that it is performing accurately. Some examples of machine learning are self-driving cars and recommendation systems, like you might see while using Netflix or shopping on Amazon.

For more resources on AI, see thetech.org/ai

AI in Action

For an example of AI, see this video about one of our 2021-2022 Tech for Global Good laureates. Wild Me uses machine learning to analyze photos, identify animals, and track wildlife populations. This use of computer vision aids scientists and researchers in their work with endangered species.

Materials 

Each team of 2-3 students will need: 

 

 

Each student will also need: 

  • Pencil or marker (1 per student) 
  • Manilla folder
  • a few pieces of tape

Preparation

  1. Plan to have learners work in groups of 2-3.
  2. Print the Game Boards.
    • You will need one Owner board per team of 2-3 students. This board will be used in both challenges.
    • Each team will also need one Challenge 1 Board.
    • For Challenge 2, each team needs 1 animal.
      • Print a variety and plan to distribute them randomly. (Students can choose from: dogs, cats, birds, rabbits, rodents+, and snakes)
  3. Assemble the boards by attaching the papers to manila folders.
    • If time permits, students can build their own boards in class.
    • Tape or staple one worksheet to the inside of each flap of the manilla folder. Both worksheets should be facing upright when the folder is open horizontally.
  4. Print the Lost Pet Cards  
    • Each team will need at least one Challenge 1 card. 
    • Each team will also need at least one Challenge 2 card.
      • Again, print a variety and plan to distribute them randomly. (Teams will need an animal that matches their Game Board.)
  5. Try the activity with other educators or kids you know. This will give you practice with the materials to be able to anticipate student questions.

 

 

Tip: If planning to play the game more than once,
we recommend laminating the materials. 

Example of an assembled Challenge 2 Game Board

Outline

Frame the Challenge
5 min total

Activate Prior Knowledge

 5 min
Challenge 1
15 min total

The Case of the Missing Pet

 10 min

Debrief

5 min
Challenge 2
25 min total
The Case of the Training Program  15 min

Debrief

10 min

 

Lesson Directions

Frame the Challenge

Activate Prior Knowledge (5 min)
  1. Start by asking students if they have ever lost a pet or know someone who has. Discuss how it feels to lose a pet and why it’s important to find them quickly.
    Guiding questions include:
    • What kind of feelings do you think people have when they lose their pet?
    • Why do you think it’s important to find our pets quickly?
      • Guide them to think about the basic needs that pets depend on their owners.
    • What are some of the first things people do when they lose their pets?
      • How long might that take?
      • Are there other ideas of how someone could find their pet quickly?
  2. Introduce the concept of artificial intelligence (AI) as a “helper” that can solve problems.
    • Explain that AI uses information and clues to find patterns and answers, just like a detective!
  3. Provide students with a basic introduction to artificial intelligence (AI) and machine learning (ML). Use the Background Information above if needed. 

Pattern Recognition: Recognizing if there is a pattern and determining the sequence.

Pattern recognition can lead to grouping, organizing, or streamlining problems for more efficient outcomes. In addition to being a core aspect of artificial intelligence, data science, and computational thinking, pattern recognition can also be seen in sports and games.

Challenge 1

The Case of the Missing Pet (10 min) 

Scenario

You have lost your pet! They could be anywhere in the neighborhood.

Luckily you have a drone that can look for it from the sky.

The drone uses AI, so you need to give the drone's machine learning program information. Make sure it has enough information about what your pet looks like to identify it and help you find it.
  1. Introduce the scenario above.
  2. Let students know that they will be working in pairs.
    • One student will be the pet owner. The pet owner will give clues about what their pet looks like. 
    • The other student is the drone. They will try to identify or "guess" which image in their dataset is the correct pet.
    • Note: In teams of 3 have students take turns either guessing or providing clues. 
  3. Similar to the game of "Guess Who" the AI in the drone can only use the information it was given to identify the pet.
    • Review the constraints for the pet owner/clue giver so that they understand the types of clues that will be the most useful.
  4. Once students understand the directions distribute the materials. Each team will need:
  1. Once students have played one time, have them switch roles so that they each have a chance to be the pet owner. 

 


Pet Owner (Clue Giver)


AI Drone (Guesser) 
  • Select a card. This will be your missing pet. Don’t share this with anyone!
  • Provide clues to the AI Drone (Guesser).
  • Confirm if they pick the correct image.
  • Based on the clue provided, select the image that best fits the description.
  • If the guess is incorrect, ask for another clue.
  • Review all of the information you have recieved and select a new guess.
Take notes of your clues on your Owner Game Board Take notes of the clues you recieve and guesses you make on your Challenge 1 Game Board

Constraints for Pet Owner/ Clue Giver:

  • Do not say the type of animal it is (e.g. dog, cat, bird, etc.)
  • Use no more than 5 words. Try to be short, and concise. 
  • Focus on what the animal looks like. For example:
    • Lines near the eyes
    • Orange and white 
  • Remember, the AI drone will not know your pet's name, or what it likes to play. The only information it has is the picture.
Debrief (5 min) 
  1. Lead a short debrief once everyone has had an opportunity to play both roles at least once.
  2. Debrief Questions can include:
    • How many clues did you need until you finally guessed which animal you were looking for?
    • What part(s) of the first clue helped you narrow down which animal you were looking for?
    • In your opinion, which clue helped you the most and why do you think it was helpful?
      • (optional) What if this was the only clue you got? How would the clue have helped you or not?
  3. Share with students that the more information (data or in this case clues) the AI system has, the better it becomes at predicting or guessing.
    • The Guesser or AI system was responding to information and data provided by the clue giver. So the more information provided, the more accurate it will become. 

Challenge 2

The Case of the Training Program (15 min) 

NEW Scenario

Petco Love Lost was so impressed that you were able to train a drone to identify your missing pet that they want to test and see if it would be possible to help them find and identify a specific animal in a group. 


For example: Could they identify a poodle if all of the other animals are dogs? 


You need to give the drone's machine learning program enough distinct information about what the animal looks like so it can be identified and reunited with their owner(s).
  1. Introduce the new scenario above.
  2. Because they are training the program, students will also be asking the AI how confident it is that the guess is correct. So, the guesser will also say:
    • very confident
    • somewhat confident or
    • not confident 
  3. The materials for Challenge 2 are divided by animal species. There are five types of animals which students can use:
    • Dogs
    • Cats
    • Birds
    • Rabbits
    • Rodents+
    • Snakes
  4. Choose animals based on student interest or the difficulty of the data-set. 
  5. Once students understand the new directions distribute the materials. Each team will need:
  6. Once students have played one time, have them switch roles so that they each have a chance to be the guesser.
    • You can also have teams switch animals and see if they can identify a different species more easily. 

Confidence Levels

AI systems usually have a confidence level to which it believes to be accurate. This is dependent on the data provided. The more data the better the confidence level it will have.

  • The game-play for Challenge 2 is the same as Challenge 1.
  • However, this time, when when the AI system provides a guess, they should say how confident they are that they are correct. 


(Clue Giver)


(Guesser) 
  • Select a card. This will be the animal you are trying to find. Don’t share this with anyone!
  • Provide clues to the AI Drone (Guesser).
  • Ask: How confident are you that it’s that animal?
  • Confirm if they pick the correct image.
  • Based on the clue provided, select the image that best fits the description.
  • Determine how confident you are that the image you selected is accurate:
    • Very confident
    • Somewhat confident
    • Not confident
  • If the guess is incorrect, ask for another clue.
  • Review all of the information you have recieved and select a new guess.
Take notes of your clues on your Owner Guess Who Board Take notes of the clues you recieve, your confidence, and the guesses you make on your Challenge 2 Guess Who Board

 

Beginning Programmers
  • If students need additional practice, model the role of clue giver for them. Project the image of the Guess Who board and have them try to guess your animal. 
  • Have begining programmers start with the dog data-set for Challenge 2. The variation in images may be easier for students to identify.
  Advanced Programmers

Have students learn more about confidence levels by using Teachable Machine. This tool has students train a computer to recognize images. Once the computer has been trained it will make a prediction on what it thinks the image is and it will provide the confidence level to which it believes to be accurate.

Debrief (10 min) 
  1. Lead a short debrief of the activity and the concepts. Help students make connections between the skills they used in this activity, the work of computer scientists, and real world applications of AI and ML.
    • When you were the “Guesser” you were playing the role of an AI system in predicting which image might be accurate based on the clues (information) that were given by the “Clue Giver.”
    • When you expressed how confident you were in that the image was accurate you were also simulating an AI system. If there is high confidence that the guess is accurate it will make the guess, if it’s not very confident in the accuracy of the guess then usually it doesn’t make the prediction.
  2. Possible Debrief Questions include:
    • How many clues did it take before you guessed the right animal?
    • What clue helped you the most?
    • What was the hardest part about guessing?
      • What was the easiest part of guessing?
    • What new things did you learn about AI and how it works?
    • Did you feel like the AI (the guesser) was learning and getting better at guessing with each clue?
    • How do you think AI could be used to help find other things besides lost pets?
    • What might go wrong if AI doesn't have enough data to make an accurate prediction? 

Standards Connections

CSTA Computer Science Standards
Grade Standard Description
K-2
 1A-DA-07

Identify and describe patterns in data visualizations, such as charts or graphs, to make predictions.

Vocabulary 

Term Definition
algorithm

Step-by-step instructions to solve a problem.
artificial intelligence (AI)

most often refers to a device or program designed to mimic aspects of human intelligence to complete complex tasks and make decisions.
confidence level How accurate or sure a computer (or AI system) is about a response or answer. The higher the level the more confident or sure the system is about the answer it provided.
data

Information that can be collected, analyzed, and used to inform decisions.
data set A collection of data
drone A small device that can fly using a remote control.
machine learning (ML) A branch of AI where the goal is to create a program that improves over time or “learns” as it processes more data.
pattern recognition Recognizing if there is a pattern and determining the sequence.

More AI Activities

The fun doesn’t stop here! AI models can be trained to identify more than just two different types of animals. Apply this process and train a model to identify other animals such as birds, dogs, horses, etc.

Check out our other artificial intelligence activities to apply critical thinking to other AI models. 

Artificial Intelligence

Visiting The Tech Interactive

Want to see artificial intelligence in action? Check out our Animaker exhibition.

Animaker uses machine learning, 3D scanning, and visual recognition to identify the animals you show it.

Buy Tickets