The Kevin Index

Providing a new perspective on ways to detect AI-generated text in academic papers. Using statistical methods to differentiate human writing and AI writing through a non-token-based method.

Comment

Inspiration

AI checkers use tokens to check if the text/paper is AI-generated, but they have been proven to not be that reliable. We decided to work on our own way to differentiate AI-generated text to solve the problem.

What it does

Our program takes in text and uses a variety of statistical measures to calculate a final index score where a more positive value indicates a human wrote it, and a negative value indicates an AI generated it. We call this value: "The Kevin Index".

How we built it

We used a Python backend and NLP libraries to parse and analyze data and output to our html front end.

Challenges we ran into

Finding good data to get a statistical baseline with.

Accomplishments that we're proud of

Being here, having fun, and doing our best.

What we learned

AI-generated text cannot be so easily detected.

What's next for The Kevin Index

More.

Share this project:

Updates