10 More AI & Machine Learning Projects for the K12 Classroom — PETE&C and ISTE Session Resources
About the presenter:
Dr. Scott Garrigan is Emeritus Professor in the Lehigh University College of Education. He has worked as K12 teacher, school district technology director, director of professional development, and university professor. He has developed courses for Lehigh University, Wilkes University, Discovery Education, and corporate clients. He offers professional development and consulting for schools and businesses across the U.S. and internationally. Contact him at:
Dr. Scott Garrigan
scott.garrigan@gmail.com
LinkedIn: Dr. Scott (never up-to-date)
+1 – 484-894-6162
Note about the projects below. None require commercial software, subscriptions, or hardware. Nearly all can be done on a laptop, desktop, or tablet. Each one is TEACHER DIRECTED. They all rely on the teacher to set the context, provide direction and support, and follow up with reflection. They can be done across many subjects and grade levels, again, according to teacher interest and skill (computer science majors not necessary!!!). All are practical for a classroom and give students insight into some aspect of data analysis, artificial intelligence, and machine learning.
Teacher note: Each project demonstrates a different aspect of “intelligence.” As appropriate, facilitate a discussion of intelligence, comparing the AI or ML system to human intelligence, starting with “Is Siri smart?” “How intelligent is Alexa?”
- Natural Language Processing (NLP) with Alexa: Alexa, what did I say? Alexa, what did you hear? What is this word in Spanish? … in Chinese? Note: this doesn’t work for Siri, who will not answer “What did I say?” Alexa “hears” within the device, but it sends the raw audio to the cloud where the meaning of the speech is determined. That result is sent back to the device to respond to our question. So Alexa “hears” very primitively with many errors. Siri, on the other hand, does all its processing in the device. So Siri should not need to be connected to the internet, but Alexa must be.
- Text Classification Project: One of the main kinds of machine learning applications is Classification. For example, a self-driving car must classify objects it sees into categories like car, truck, bus, school bus, bicycle, pedestrian. Chatbots and digital assistants much classify words or text into categories like requests, commands, choices — and then into narrower categories. uClassify.com is a free site offering nearly 100 pre-trained text classification examples. Students can create their own text classification projects using the pre-trained models as examples. To train an original, private text classifier, an e-mail address is needed to create a free account and to receive private “API codes” that allow your text classifier to used in a real chatbot or other AI application.
- “Rock-Paper-Scissors game” Image Classification Project: There are two image classification sites designed for young children (ages 7 to 13) that use custom machine learning extensions for Scratch. One is an MIT research project: https://cognimates.me. There’s a very cool activity to train a machine learner to recognize live camera images for a game of rock, paper, scissors. There’s a video on the site to show the image classification game in action. The other is an awesome one-man project from Dale Lane, a UK educator who is passionate about introducing kids to machine learning. Here’s a page that shows activities he did with a group of 8-14 year olds that also included the rock, paper, scissors activity: https://dalelane.co.uk/blog/?p=3537 . The 20-minute video on his main page clearly presents what his “MachineLearningForKids” activities are like: https://dalelane.co.uk/blog/?p=3524 NOTE: both of these sites may have added complexity that would make the activities difficult for a whole class with one teacher. Breaking the class into 3-person teams or having a tech support person on hand can make it manageable.
- Build a Chatbot: The free chatbot service https://snatchbot.me
Build your own chatbot: https://www.chatbot.com/help/build-your-chatbot/how-to-build-your-chatbot/ You, your students, or parent needs to create a free account that requires a valid e-mail to use this free site. - Explore a Genetic Algorithm with Genetic Walker. Younger students (4th to 6th grade) can watch the genetic walkers slowly learn to walk; older students (7th to 12th) can modify genetic parameters like gene mutation probability and gene mutation amount. Fitness scores on each mutant in each generation are displayed, and the top scorers are recorded to document species evolution. The HTML5 animation should run on any device. Here’s the site: https://rednuht.org/genetic_walkers/ After a day of learning from random mutations, see the progress here: https://www.youtube.com/watch?v=XiJ537K_9ww
Interested high school students can learn more about genetic algorithms built on random mutation that underlie Genetic Walker with this 12-minute video: https://www.youtube.com/watch?v=uQj5UNhCPuo
- Easiest introduction to “R.” The computer languages used for data analysis (R) and machine learning (Python) may not be taught in your school, but you don’t need to know the language well to run and modify data and machine learning models. Many programming languages don’t run on a tablet or chromebook, but this example runs in a browser “notebook,” so it should be device agnostic. Here’s the SIMPLEST introduction to “R” running in a friendly notebook environment. The 16-minute video gives narrated, step-by-step instruction for students to enter and run simple R beginner commands: https://www.youtube.com/watch?v=rUJiolRoH1M
- Classic Titanic dataset classification activity using a Random Forest algorithm in R (8th to 12th grade). Both the dataset and R script are found in Kaggle, and the activity can be run in the Kaggle Notebook. Here’s a link to the activity: https://www.kaggle.com/vincentlugat/titanic-data-analysis-rf-prediction-0-81818. Here is 25-minute narrated tutorial for a step-by-step explanation of the code (the speaker is slow so you can follow along. Speed up video as you wish): https://www.youtube.com/watch?v=icAJ3wVxyik
- OpenML for High school computer science or programming courses: OpenML.org is dedicated to providing a friendly, resource rich environment to study Machine Learning. Their site includes 21,000 datasets and thousands of strategies and code examples for analysis and prediction in topics ranging from credit scores and tic-tac-toe to blood analysis. There’s a LOT in this site that is accessible, mostly for intermediate learners. Watch the 1-minute video introduction to OpenML to sense the richness and accessibility of the site: https://www.youtube.com/embed/1N3qATxXrpE
Here’s the OpenML site: https://www.openml.org/ - Build a Recommender System. This is a bit more advanced, so high school juniors, seniors, and other very interested students. The activity page has both sample code as well as lengthy explanations and good diagrams. It assumes you’ll run Python in a very simple environment, so there may be small differences in the program setup you’ll have to figure out (like loading the dataset). Here’s the page: https://towardsdatascience.com/how-to-build-a-recommendation-engine-quick-and-simple-aec8c71a823e
- Explore OpenAI’s Image and Text generation with DALL-E and ChatGPT (GPT-3): The creative art application DALL-E rocked the internet world in 2022 and the text dialog program ChatGPT in 2023 became the fastest-adopted program in history. You need to create a free account for both of these which can be your existing Google account. Here’s the DALL-E page with examples and a button on the bottom to create your account and begin your art exploration: https://openai.com/dall-e-2/
GPT-3 and ChatGPT are “Large Language Models” designed to create human-like conversation and narrative. These models are trained on language fluency, but they have no real-world experience or deep subject knowledge. They are particularly weak in mathematics. As of early 2023, the models completed their training in June, 2021, so they don’t know anything that has happened since then. To explore GPT-3 (on which ChatGPT was based), click the Getting Started link in OpenAI’s section on GPT-3: https://openai.com/api/
To continue experimenting with GPT-3 try the Playground for longer text experiments. Note that the site can be extremely busy, so it’s sometimes inaccessible: https://platform.openai.com/playground
ChatGPT is so popular, its servers are often too busy to access, but you can try here (you’ll need your OpenAI password, and it’s free). Learn about how to create a meaningful “prompt” before wasting time in uninformed exploration. It’s very powerful. The link: https://chat.openai.com/chat