GradeAI

Making essay grading easier through AI, saving teachers time!

Comment 3

Slideshow

Link here: https://www.beautiful.ai/player/-N_1SnNYS3rJPTXJv7Dy/Untitled-4

Introduction

Teachers bear the responsibility of guiding and inspiring young minds, navigating the challenges of limited resources, diverse student needs, and the relentless pursuit of fostering knowledge and growth. Yet out of all, teachers have to spend a lot of time grading, which can cause a serious toll on their sanity. So, we at GradeAI created an application that directly addresses this issue: GradeAI not only grades essays with great accuracy but also grades efficiently. Instead of taking 2+ hours grading simultaneously, teachers can effortlessly grade their student's papers and focus more on things with more meaning!

What does GradeAI do?

GradeAI can do a plethora of things. Let's take a look at some of the important ones:

  1. GradeAI can quickly assign a grade to an essay according to the course's expectations, rubric, and assignment prompt. The consistency of the grades allows teacher to completely eliminate bias within their grading.
  2. GradeAI can give feedback on each paper to explain why each essay has the grade it was assigned. This not only allows the teacher to understand the limitations of the students themselves but can give them insights on how to teach their students better.
  3. GradeAI can accept many formats of essays, including text, pdfs & other readable files, and also images. With the power of text recognition and OCR models, GradeAI has flexibility when it comes to accepting essay formats.

How did we build GradeAI?

We used Python and Flask primarily for the back-end part of GradeAI, and with the combination of text-recognition models and GPTs (generative pre-trained transformers), we could connect the front-end with inputs (files, text, etc.) and outputs (grades, feedback, etc.) in an efficient manner. The front-end was made with HTML/CSS, JS, TS, React.js, & Next.js. This, of course, came with some problems (one being inaccurate AI results) but we smoothed them out along the process.

Built With

Share this project:

Updates