TutorAI

Inspiration

TutorAI was inspired by the realization that, despite the fact that most people enjoy learning, many of us struggle to find resources tailored to our classes or topics of interest. We realized that with the advancements in technology, specifically the interpretive capabilities of the GPT software, we could create a platform that would make learning easier and more accessible for everyone. Our goal with TutorAI is to democratize education and break down the barriers that prevent many people from learning new things. We aim to do this by streamlining the education process, making it more efficient and convenient. With TutorAI, you can learn anything, anywhere, at any time.

What it does

Essentially, our application is meant to serve as a customizable private tutor. With a Youtube video as input, the text-generative AI can help you digest the material of the video with the snap of a finger with a suite of different tools: concise and accurate automatic note-taking, a Q&A service that directly cites lines from the video, and a quizzing functionality to test your knowledge.

How we built it

The project utilizes a Python Flask framework backend that uses the OpenAI Whisper and GPT-3 API’s to transcribe and interpret the video provided, directly pushing the information requested to the display. The frontend is built with the ReactJS framework.

Challenges we ran into

This was our first hackathon as sophomores and juniors in high school, and we ran into some challenges with using React JS for the website since we were not very familiar with it. We were able to learn the language with the help of mentors and tutorials to develop the front end. We also had trouble connecting our backend and frontend, but with some persistence we managed to sync the two stacks. Additionally, it took time to find the right prompts to ensure that OpenAI’s models returned the correct information.

Accomplishments that we're proud of

As high school students we are proud to have completed our first hackathon. Not only did we learn a lot, but we were able to complete a project in the AI/ML domain that we think could deliver real value to learners around the world.

What we learned

We learned about many different frameworks and APIs in the AI/ML space, such as React JS, the OpenAI API, Flask, and JSON. We also learned how to work efficiently as a team and persevere in difficult times. We discovered how to build an easy UI that users can seamlessly interface with. Most importantly, we learned that, if we work together, we can accomplish anything!

What's next for TutorAI

We used the GPT 3 model on transcripts from lecture videos, but in the future we hope to be able to use OCR to detect text from boards and slides in videos to finetune the model. We also hope to add direct file input rather than just the URL. We also want TutorAI to reference specific timestamps in the video and allow the user to seek directly to the relevant part in the lecture.