We’ve all seen the cartoons and movies with robots that talk and act like humans (e.g., The Jetsons, Iron Man, The Terminator). Sometimes they are scary and sometimes they are helpful. In any iteration of the idea, artificial intelligence (AI) can often seem like a distant, futuristic dream, but what if it was real?
Machine learning and AI are more and more relevant in today’s technology, but they are not necessarily the same thing. How do we tell the difference between them? How are they used in society right now? Why are they so helpful? And how do we want to see them grow in the future? In this activity, we will learn about machine learning and how it is part of the wave of artificial intelligence. We will train our own machine to recognize images and brainstorm ways we could use this to solve problems around the world! For more in depth information refer to Appendix A.
Developed by Actua’s Network Member Science Venture
To Do in Advance
- For this activity, we will train a computer to recognize text or images, and associate it with an output. We will experiment with how accurate its outputs are, then discuss ways of possibly improving them.
- To complete this activity you will have to set up a class account with Machine Learning for Kids and access the teacher portal (free) to set up your API keys and class accounts.
- Here is a detailed Step-by-step guide for creating class accounts. Follow the instructions and review the different options to figure out which class type would work best for your group.
- Note: you are only able to have two machine learning models with each free API key. However, As educational institutions, we can apply to access the paid version for free. Instructions for doing so are also included in the step-by-step guide.
- After your API keys are set up, proceed to creating the participant accounts and create a system for every participant to access their account. For example, assign each participant a number and explain that their account is SV plus their assigned number. The password would be the same for every participant.
Opening Hook: Introduction to AI and Machine Learning
- Using a smartphone or other device with home assistance (e.g., Amazon Alexa, Google Home, etc.) – interact, for example: “Hey Siri… tell me a joke for kids” (instructors can preview what Siri will say before the phone or computer speaks)
- Have a discussion of what they think it means to be intelligent. What can people do that machines can’t? What can machines do that people can’t? Computers at their core are just things that compute.
- Here is a cool example of some AI at work:
- But why are we so obsessed with AI, and how are we already utilizing the technology in our daily lives? Is there AI in our phones? In social media? Discuss the differences between AI and Machine Learning.
- AI is a program that makes a computer act as if it were intelligent and adapting to changing situations, but it was all designed by a human. Machine Learning is when a computer will alter itself based on a sample set of examples given to it by a human, and works by making predictions based on patterns it finds in the data. In this way, computer intelligence can be pre-planned or “trained” through experience.
Section 1: Training a Machine
- As a teacher, you will be able to co-create the “labels.” These are the names assigned to the objects the machine will recognize and associate with the data points. Navigate the project page to select your project and begin creating labels.
- Examples of possible labels include;
- Black and white,
- Cups, bowls, and forks,
- And Pens, pencils, and markers.
NOTE: The more labels you have, and more data points under each label will increase the time it takes for it to train. We suggest keeping it to two labels and fifteen data points for each to keep the time it takes to train the machine below ten minutes. However, if you wish for a more accurate machine, you will need to input many data points and be prepared to wait for it to complete its training.
- Ask the participants to find objects around the classroom that match the labels you have co-created and use the webcam option to take photos. These photos become your data points that the machine will use to train itself. Alternatively, the data points could be obtained from the internet or uploaded from images saved in the computer.
- Depending on the age group it would be more or less challenging for the youth to find images online. For younger groups you could have the data points already uploaded, while the older participants can either create or find their own data points for the machine.
- Once you have at least ten photos under each label, back out and select “Learn & Test.” From here, the computer will analyze the data sets, and in a short period of time, will be able to deliver feedback when presented with new images.
- While waiting, consider having kids toy around with Google Quick Draw here.
- Once the process is complete, you will be able to submit photos of items that you would consider to fit under one of the labels and see what the machine says.
Appendix A: Background Information
Overview of AI and Accompanying Terms
- The differences between AI and Machine Learning
- AI and machine learning as available technology
- AI and machine learning as helpful tools
- Artificial Intelligence (AI), the theory and development of computer systems able to perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.
- Strong AI: a machine with consciousness, sentience and mind) or artificial general intelligence (a machine with the ability to apply intelligence to any problem, rather than just one specific problem).
- Weak/Narrow AI: focused on one narrow task
- Type I: Reactive AI: Perceives its environment/situation directly and acts on what it sees
- Type II: Limited Memory: considers pieces of past information and adds them to preprogrammed representations of the world
- Type III: Theory of Mind: capacity to understand the thoughts and emotions which affect human behavior. Comprehends feelings, motives, intentions, and expectations, and can interact socially
- Type IV: Self Aware: extension of type III, aware of internal states, can predict the feelings of others, and can make abstractions and inferences. Super intelligent, sentient, and conscience.
- Deep Learning, part of a broader family of machine learning methods based on learning data representations, as opposed to task-specific algorithms. Learning can be supervised, semi-supervised or unsupervised.
- Machine Learning, is a field of computer science that uses statistical techniques to give computer systems the ability to “learn” (e.g., progressively improve performance on a specific task) with data, without being explicitly programmed
- Supervised: machine learning task of learning a function that maps an input to an output based on example input-output pairs.
- Unsupervised: machine learning task of inferring a function that describes the structure of “unlabeled” data (i.e. data that has not been classified or categorized).
- Artificial Neural Networks, computing systems vaguely inspired by the biological neural networks that constitute animal brains.Such systems “learn” to perform tasks by considering examples, generally without being programmed with any task-specific rules.
Reflection & Debrief
- What are things that would lower the accuracy of this machine learning model?
- The machine learning model
- Not having enough high quality examples
- Camera quality of the computers
- Light quality in the room
- How could we improve our machine learning model?
- Having more data points! – The more data the Machine has to learn from allows the Machine to observe an increased amount of diversity while still knowing that the data point is “x.” This increase in variety enables the Machine to be able to judge a broader set of data with an increased amount of certainty.
- How has machine learning benefited us in our daily life?
- What could we create knowing what we know now?
Extensions & Modifications
How might you adapt the time, space, materials, group sizes, or instructions to make this activity more approachable or more challenging?
- Add on a Scratch activity that focuses on machine learning to integrate beginner level coding into the workshop. “Make me Happy” is a simple beginner level Scratch lesson plan in which participants teach the computer to recognise whether they are insulting it or giving it a compliment. This extension will add approximately 45 minutes onto the original workshop.
- For younger groups you can already have the data points loaded onto the machine and have it trained. The younger participants could just learn about machine learning and test the model – it could be turned into a game. This could be done with drawings, colors etc.