top of page

Learnings from 100,000 AIs built by K-12



I recently discovered that AIClub students have built nearly 100,000 AI models! What have we learned from this experience? What can other educators do to bring AI literacy to their classrooms?


How did we do this?


AIClub has been teaching students artificial intelligence since 2019, when we were one of the first to do so. Every year students working with us build more and more AIs. We both teach classes ourselves and work with teachers to bring AI into their classrooms. This combination has offered us a unique perspective on how to create effective AI learning experiences for kids of all ages.


One size fits all? No!


AI enters a world where knowledge of computer science, data literacy, and technology availability varies widely. We have found that a range of curricula and tools are required to meet kids and teachers where they are, and take them where they want to go. These include a range of no-code tools to full code exercises coupled with a range of problems from easy exercises that are fun to meaningful problems with depth to explore. We have created Navigator that enables students to build AIs with no code, tune AIs, connect to the AIs with code, and learn how to code the AIs fully end to end when they are ready. This seamless path has helped many students start early, create, modify and learn.


Is code needed? Maybe!


The next discovery we had was that coding ability varies dramatically. We frequently encounter students with every possible coding skill level (beginner to expert) in every grade level from 5th upwards. Creating curricula that work with every level of code, from no-code to advanced python, helped us meet this breath of incoming knowledge base.


Engagement


One item that all of our programs have in common is custom projects. Every student creates a custom project, matching their unique interests with their AI learning.


Examples of simple projects

- Use AI to create art!

- Learn about AI Bias using famous datasets like COMPASS

- Teach an AI about an animal, or a planet, or about pokemon!


Examples of deeper projects

- Use AI to detect toxic speech, bias, or deep fakes

- Use AI to detect or forecast disease

- Use AI to mitigate climate change.


Over the last 3 years - we have found that this approach generates a unique engagement between the students and the technology. They are proud of their creation, engaged and inquisitive, and driven to improve what they have built. Along the way they gain a practical and realistic assessment of the technology and their role as humans in creating solutions that combine their ingenuity and the technology's capabilities. Our student projects have won over 130 national and international awards in everything from science fairs to prestigious professional publications. Examples of student projects - in their own words, can be found here.


The multi-disciplinary aspect of AI also means that simple projects can be integrated into any classroom - math, science, literature, etc.


Diversity, Equity, and Inclusion


We have seen positive results of AI engaging all students. Last year, we released our diversity study. We have found that a far larger percentage of our students are girls than is the norm in computer science. We have also found that over 50% of our competition wins are by girls, demonstrating that girls not only enjoy learning the technology but are driven to use it to reach new heights. See our diversity report here.


The light in a student's eye


For me, the greatest joy in this experience is the light in a student's eye when they see their AIs work, and see their effort in training and tuning their AIs rewarded. Through these 100,000 AIs, we have seen countless students develop a passion for technology and STEM, grow their own confidence, see their imagination come to life, and develop a practical and realistic understanding of their role as humans to guide and shape this foundational technology for their generation.


90 views0 comments

Comments


bottom of page