ORANGE (Output)

The following activity will help you practice identifying the data around you and the potential features and labels an AI system might use to determine an output.

Activity 7: Data and You

Using the blank tables in your workbook, choose an object, person or phenomena relevant to you or your participants. Determine what features and labels you would use to train an AI system and determine what the input and output would be for each example. Refer to the previous apples and oranges example if you need more direction. This activity is a great way to introduce the idea of labelling data to your participants!

Example: _______________

Input: 

Output:

Example: _______________

Input: 

Output: 

Reflection Questions

After completing the data tables, consider the following questions. You may use your Activity Workbook to track your thoughts.

• Why are features and labels important to training an AI?
• What are other relevant examples of data that your participants may be familiar with? 
• How would you modify or extend this activity for various age groups of participants?

Machine Learning

Machine learning (ML) is a branch of artificial intelligence (AI). It focuses on programming computers and machines to imitate the way that humans learn, to perform tasks autonomously, and to improve their performance and accuracy with more experiences or by analyzing more data.

Adapted from: https://www.edureka.co/blog/machine-learning-tutorial/

Above is a visual representation of machine learning with each step explored below.

1. Data is fed into a machine learning algorithm so the data can be first analyzed by an expert and labelled. As we explored earlier, labels are tags that are associated with the data, such as the type of fruit, fish, career advice, results of a test or implications of using a certain teaching technique. 
2. The machine learning algorithm then learns the trends of the data and builds a model of the data. This step is known as training the machine. It could be thought of as training someone to do a task and then giving feedback about whether they have been successful or not.
3. The trained algorithm can then be used to predict or identify items based on its training. In the figure above, the algorithm is trained on apples and the supervisor is telling the algorithm that it is seeing apples. Once the algorithm is trained it is then used to recognize an apple.

A more in-depth explanation of some of the different types of machine learning can be found in the Additional Learnings section of this module.

Generative AI

• Generative AI also uses machine learning but is specifically designed to produce or create content based on its dataset.
• Large Language Models (LLMs), like ChatGPT, are given massive amounts of data and are trained to understand the patterns in human speech. This allows us to interact with them as if we were interacting with a human.
• Image generators work in a similar fashion but instead of processing words they are given large amounts of image data with long lists of features to help it create a new image based on a prompt. Once again the AI looks for patterns it can learn from.
• Music generators and voice generators analyze the audio data in a piece of speech or music and learn patterns from the dataset.

Machine learning has a multitude of applications across all aspects of daily life, industries and STEM disciplines as showcased below. Consider which of these applications may be relevant to your participants.

Image from: https://blog.quantinsti.com/machine-learning-basics/

Exploring AI Applications

AI applications have been around for longer than most people think. Apple’s Siri and a traditional GPS are both examples of early AI technology. As machine learning systems have improved, what AI can do has become far more efficient and impressive. From advanced Large Language Models (LLMs) like ChatGPT to self-driving cars, the applications of AI in our world have become far-reaching. The following activity will introduce you to a few applications of AI. 

For a curated list of AI applications for educational settings refer to the Future Ready Section.

Activity 8: AI Exploration Stations

Explore the following stations by following the steps below. After exploring each station, consider the following questions. You may use your Activity Workbook to track your thoughts:

• What new skills or knowledge have you gained?
• What challenges did you face?
• How could this be incorporated in different programming types (camps, clubs, workshops) and for different ages of youth?
• What part(s) of the Substitution, Augmentation, Modification, Redefinition (SAMR) Model could these tools address?  

AI in Society 

It becomes increasingly important for participants to understand not only how AI systems work but also how they are used in their day to day lives. Examples of important questions that both you and participants should be asking yourselves about AI could be:

• Where was the data for this system sourced from? Was it consensual?
• Is there an environmental cost associated with the growing use of AI?
• Is AI safe to use? What about AI systems that could put people in dangerous situations like self-driving cars?
• Who is responsible for any negative consequences from using AI? Is it the creator, the user, both?

Weaving opportunities for reflective discussion about AI and its use in society can really help to bring relevance to your content and programs. Removing the abstraction from AI and showing participants its use in their lives helps keep students engaged with the topic.

Video: We’re already using AI more than we realize  (Vox, 0:00s-6:31s)

Activity 9: EthicAll Dilemmas

Part 1

First, play  the video, We’re Already Using AI More Than we Realize (Vox, 0:00s-6:31s). Then, choose one use of AI that you think you or your participants use in their daily lives. Examples could include:

• GPS Apps
• Smart Home Systems (Alexa, Siri, Google Assistant)
• Recommendation Algorithms (Instagram, Youtube, Spotify, Netflix, etc.)

Reflection Questions: 

• How do you think this AI system works?
• What kind of data does it use to make decisions? Where do you think it comes from?
• What ethical implications does an AI like this carry? Why?

Part 2

Explore MIT’s Moral Machine website. This project was designed to crowdsource data related to the way humans make ethical decisions regarding life and death. Work through the following steps:

1. Click on the “Start Judging” button and choose an option for each of the 13 scenarios.
2. When finished, analyze your results. Do they reflect your own morals?
3. Reflect on the implications of a self-driving car being trained by someone whose morals differ from yours, as their biases will inevitably influence the AI they develop.
   • What are other examples of AI systems that may have problematic biases?

Now that you’ve had a chance to explore AI, data, and how they work together, explore the resources we’ve developed in the Developed Resources section to continue to learn and explore!