AutoMathic

Inspiration

With the rise of AI advancements, academic integrity is at an all time low. AI is a tool with the potential to empower students, but many students abuse it as a crutch, which ultimately is not conducive to effective learning. We sought to create an AI tool that empowered students' learning without overpowering them.

What it does

Our website is a digital notebook which checks students' mathematical work line by line, letting them know when they make a mistake. We help students unlearn bad habits by immediately delivering feedback, without giving them too much assistance. Automathic prefers to take a backseat and let the student handle most of the work, while making sure that they stay on the right track. We also created an automatic assignment feedback system where students can submit their homework assignments and receive feedback on it before they go to the instructor's office hours.

How we built it

We used MongoDB to store assignments, student accounts, and teacher accounts. The backend was created with express, and runs on Node. For the equation recognition, we used a 3rd party convolutional neural network which took in an image and returned LaTeX formatted equations. Using these equations, we parsed them with Mathematica and checked for equality at each step of the way. For the frontend, we used TailwindCSS for the styling and Next.js/React.js.

Challenges we ran into

One challenge we ran into was Javascript's inferred typing. We had a really hard time debugging numerous bugs during the hackathon where we thought we were supposed to be passing in an image, but we were really supposed to be passing in that image's bytes.

Accomplishments that we're proud of

This was our first time working with AI and we were pretty happy with the end result. At the end of the day, we made a product that we would actually be able to use in our everyday lives.

What we learned

We learned a lot about AI and although we initially thought it would be finnicky and undependable, we quickly began to trust and depend on the results of the model.