Inspiration
We were inspired by the efforts of teachers in helping their students grow both in their studies and as people. However, we noticed the gap that existed between teachers and students, especially after the acceleration of technological advancement. Now, with the prevalence of AI, students can more easily feign their knowledge, which hinders the learning process, as the student is merely given the answer without understanding and the teacher doesn't know in what way to assist the students. We want to help facilitate a closure of this gap by using newer technology to fill in this missing piece, turning something students often use into a tool to further growth and learning.
What it does
Our website consists of student and teacher accounts, with teachers having classes that they can bring students into. Our website allows teachers to assign problems to students in their class and have an AI assistant to help guide students in how to do said problems. We also track questions asked by students in order to get a better idea of what topics specifically students are struggling with. Finally, we provide that information to the teacher to inform them on what topics students need more help in.
How we built it
In terms of backend, we had two major sections of our website function; one side generated responses to guide a student in how to do a problem and the other compiled and analyzed student data for teachers.
We used OpenAI to generate responses to the student asking questions on the assignment. We particularly paid attention to making sure our AI guides the student to the answer rather than immediately telling them the answer.
When providing student data for teachers, we first compiled student data using questions they asked while while students complete assignments. and when we're done, we pass it through AWS Claude to classify those questions. Using these classifications, we can see what topics students are struggling on the most as well as which topics each particular student needs more help on. We can then store this data with its associated classes, teachers, and students to be accessed later.
For our frontend, we chose a colorful background to create an engaging atmosphere for students. In particular, we thought the changing colors may bring more wonder in relation to learning. Beyond that, we chose to focus on functionality to make our website as straightforward as possible for the user, both teacher and student, to use.
Challenges we ran into
One of the first major challenges we ran into was that one of our features hinged on an API that we ended up being unable to use. As a result, we had to piece together multiple API's to create the same effect: converting a pdf file to usable text.
Another major challenge we encountered was creating our database to store teacher, student, and class information. It was confusing to thing about the best way to format our data structures through MongoDB and we had to think ahead to what we'll have to access when considering the database format.
A third challenge we encountered was a lot of our API's being inconsistent to access. We frequently had to pause and figure out why our API keys or connections weren't working, which slowed down our testing process significantly. Still, we managed to overcome this by testing with hard coded values before drawing data from a connection.
Accomplishments that we're proud of
Something we're proud of is that we created a UI that aligned with our vision for our website. By this, we mean that the different tabs lead into each other in the way that we had wanted. Another accomplishment that we are proud of is how we learned to use more than one model to suit our needs.
What we learned
We learned a lot about queries for LLMs, as well as different ways to structure databases. Especially because we had encountered a lot of challenges in how our database was structured and with our queries to LLMs, we different techniques we could use to minimize the negative effects or eventually get to where we want.
What's next for TAI
There is a lot in the future for TAI, including more accurate reports on student developments and allowing for speech queries, as it could allow for more accessibility to our tool. This would make our website both more useful, but also provide benefits to a wider range of people.
Log in or sign up for Devpost to join the conversation.