Artificial Intelligence (AI) is in the news given the popularity of ChatGPT and other new products. AI can mean everything from classifying between dog and cat to writing a Shakespeare-style poem, and from crafting a piece of music to generating whimsical drawings.
Currently, learning AI is mostly limited to college CS curriculum given the difficulty behind its theory and coding. At Breakout Mentors students are able to learn AI as early as middle school by focusing on more hands-on projects with less theory! For three years now we have offered the Machine Learning and Artificial Intelligence Academy for our advanced students. Through our summer AI/ML Deep Dives, we touch the fundamentals and swim across the neural networks, and finally make amazing projects.
Let’s take a look at some of the amazing accomplishments by our students this year!
8th Grader Exploring Game Development and Music
Joseph is an incoming 8th grader and a seasoned Breakout Mentors student, having been with us for an impressive 5 years now! He has a firm grasp of coding and CS fundamentals, evidenced by his recent completion of a large video game development project. Plus Joseph is even producing his own music for the game!
When Joseph was first introduced to our AI curriculum, his shyness was apparent. We love starting with icebreakers to ensure students like Joseph feel comfortable and encouraged to interact in the learning environment. Recognizing the complexity of advanced ML concepts for middle school students, we use metaphors and analogies, such as comparing gradient descent to walking down a mountain. Joseph enjoyed the hands-on sessions that didn’t get bogged down in too much theory.
10th Grader who passed USACO Bronze in 7th Grade
Aditya is an incoming 10th grader with an advanced background in Computer Science. He passed USACO Bronze on his own in 7th grade and has been studying USACO Silver for the last year with our mentors.
Quite new to AI/ML, we were unsure of his experience coding in Python, so we began by assessing his Python familiarity and briefly reviewing the language. Aditya proved to be a fast learner and an independent thinker, allowing us to personalize the curriculum based on his pace of learning. He completed the core curriculum within just 6 sessions, not the usual 10! He then became interested in designing and implementing a Reinforcement Learning (RL) agent to play the game 2048. We introduced him to some RL concepts in HuggingFace and allowed him to explore possibilities in StableBaseline3, a tool containing the necessary algorithms. By the end of the summer, Aditya had trained an AI agent that he could watch as it “learned” the best strategies for the game!
Curious 10th Grader who Wants to Model Wildfires
Jordan is an incoming 10th grader who has just finished AP CS A. Although he had dabbled in AI/ML in the past, including a project modeling wildfires, he didn’t fully understand the concepts and code involved. Eager to learn the fundamentals of AI/ML from the ground up, he approached his learning with curiosity and enthusiasm.
Jordan asked many questions, prompting us to expand the scope of the curriculum to satisfy his curiosity. We tailored the curriculum to emphasize fundamentals, ensuring he had a solid comprehension of the material before embarking on independent coding tasks. Starting with nearly no experience and confusion about how ML models work, Jordan achieved an impressive final result. Beyond covering the fundamental concepts and theory, we explored how it could be applied to his wildfire project, weaving theoretical understanding with real-world application.
Incoming College Freshman with a Strong CS Background
Sahil is an incoming freshman in college with a very strong background in Computer Science. He has been studying data structures and algorithms for USACO and participated in an AI/ML Deep Dive with us last summer. His expressed interests include Chess and Poker, subjects that spurred his curiosity further.
We began by reviewing the Python basics and AI tools he learned the previous summer. Since he had already taken AP Calculus, our introduction to some ML topics delved more in-depth into the mathematical theory behind them, such as explaining how gradient descent works in terms of calculus concepts. Sahil grasped the concepts quickly and demonstrated outstanding coding skills. Toward the end of the curriculum, we introduced Reinforcement Learning (RL) to help him develop algorithms for playing Chess and Poker. He also built a simple AI to play the game Space Invaders. His learning journey involved not just grasping the technicalities but also fostering a deeper understanding of the theory and applications of ML.
Are you next?
How is Breakout Mentors, especially the AI Deep Dive, different from a regular CS class? First, the curriculum is personalized. These examples show the range from an 8th grader just learning the high-level basics with hands-on projects to a college-bound student diving into the theory. Second, for all the students, it is hands-on and project-based. Our curriculum is crafted toward the goal of learning by doing, which there will be a lot of hands-on activities.
You might ask, what’s next after this ML academy? One of the things that your son or daughter will get out of it is practical experience with AI/ML. They will begin to see the world is full of data and endless possibilities for AI projects. Some of these students continue on during the school year to make more advanced projects. Others will turn their studies elsewhere for the school year and revisit AI in the future. But we know these deep learning experiences have a huge impact on each student!