Groupthink

Inspiration

Online learning has always been a student-led endeavor. They are the ones leading the search for content, articles, and summaries based on what they want to learn and their experience levels. With all the content avaliable on the internet, it can be overwhelming for people to find high-quality content that is up to date and actually suits their needs. To fix this, we aimed to reverse the direction of learning: instead of the student finding the content, the content finds the student.

What it does

Groupthink is a personalized education platform that curates and delivers educational content to students in bite-sized courses automatically processed into a curriculum determined by the student's needs. When first visiting the platform, they fill out a form that includes what they want to learn and their background. The platform will then automatically generate guiding questions that help the user target and tailor their experience.

The content is then broken down into a sequential curriculum that separates it into groups based on difficulty and topic. Each of these topics then contain curated content that includes summaries and outside resources that were curated by our system based on the user's desired experience.

From experienced subject matter experts looking to specalize in a certain area, to school-aged children wanting to explore their curiosity in a field, the platform supports it through its personalized content curation and delivery system.

How we built it

The platform was built using React and the standard web framework, including HTML, CSS, and JavaScript. The OpenAI API was used for personalization and form question generation, although it was combined with our own internal parsing modules and external APIs to create the full personalization system. We also used the YouTube API to search for external videos relevant to the user.

Challenges we ran into

Due to the lack of documentation and unreliability with the APIs we used, we faced many bottlenecks in our development. Since the platform required synergy between our parsing modules and the API responses, this slowed us down a lot, forcing us to pivot and let go of features we wanted to pursue.

However, we tackled this by prioritizing learning about the APIs we needed to use and ensure that we stopped any major flaws in our usage before they happened. While we had to spend less time on development, this helped us reach a working product quicker.

Accomplishments that we're proud of

We were all very happy with how we were able to parse and transform language model outputs into a curated educational experience that went beyond walls of text. While getting the APIs to work was cool, being able to truly utilize it and incorporate it into our product took it to the next level.

What we learned

We all grew in our knowledge of React and its major concepts, including state and hooks. We also learned more about how LLMs work, especially from a programmatic context where we needed to transform language into consistently formatted text that could be parsed by our platform.

What's next for GroupThink

We want to take our core curation and expand it by making the platform more social. This includes course sharing, profiles, leaderboards, and even collaborative Q&A and 'Kahoot'-style questions that get users to help each other beyond their personalized experience.