Learn with Herm

Homepage
Users uploads files such as their course content and other resources. Using AI (ChatGPT API), LWH analyses the content + creates questions
Map. Each planet represents a topic the user can navigate to. Each topic contains questions and relevant AI-generate notes for them to study
Quiz
Score

Inspiration

Education and quality learning opportunities is a necessity for children to foster growth and empower people to be able to live sustainable lives. In order to learn topics, studying is crucial.

Students may find difficulties studying due to the following reasons: Lack of motivation Struggling with time management or may not find the time to study

One of the most effective study techniques is active recall which is actively stimulating the brain for information. This retrieval practice helps save information in long-term memory and identify topics that require more attention. Creating questions based on the material, testing yourself, and using flashcards are prime examples of active recall which forces your brain to learn the material by actively learning instead of passively reading.

We wanted to create a solution that will incentivize users to study and improve their experience learning topics. Gamifying the process encourages users of all ages to do so while improving their attention span. Through personal experiences and observations we noticed that while the active recall methods above are effective, creating the questions, notes, or flashcards may be time-consuming, tedious, and a lengthy process. As a result, we strive to minimize that time and automate the process using artificial intelligence.

What it does

Learn with Herm is a video game centered around helping students study effectively. Users can upload the content they wish to learn. Using AI-powered learning models such as the ChatGPT API, LWH analyzes the content, extracts the most important concepts, and auto-creates questions the user can use to quiz themselves. Instead of spending time generating questions or making flashcards/ notes, the user is able to allocate their time to actively learning and studying it.

When you first click into the app, you are introduced to your virtual assistant called Herm, who guides you through the game. He prompts you to add a pdf of your course notes, and then analyzes it, building a new “planet” under the subject category and turning course notes into cue cards with 4 multiple choice answers. By hitting next, you are taken to your home screen, where your past study quizzes/notes are. By hovering over a planet, you indicate which subject you want to brush up on, and can click play. Then the actual game begins, with a series of cue cards that test your knowledge of the subject and track your score.

How we built it

Game development was done through Unity using C#.
AI analysis used the ChatGPT API
Procreate and Canva for the visual elements and assets
Microsoft Visual Studio/ VS Code IDE

Challenges we ran into

The learning curve of using Unity and game development with no prior experience. Brainstorming a creative idea that is impactful and viable.

Accomplishments that we're proud of

Being all beginners to game development, we’re of how quickly we picked up a new framework and how we were able to get a working game going with that framework.

What we learned

We learned a lot about Unity through this project, including basic concepts like camera perspective, sprites, tiles, program structuring, and scripts using c#.

What's next for Learn with Herm

Learn with Herm has a lot of next steps that were not able to be implemented in such a short period of time

A more intricate storyline. For more interactive gameplay, cue card screens will have bosses that you fight by answering questions correctly
Direct connection to the ChatGPT API, which we were not able to implement due to the lack of free tokens on our accounts. Given the time and money, a direct connection would allow for more seamless creation of quizzes
A history of quiz scores and analytics to see your best subjects and subjects you need to work on
Ability to star questions that the user gets wrong and automatically generate a “subject review” test for all the challenging problems a user should review
Earn points for rewards based on outcome and using it