For my Natural Language Processing class, my group’s final result is to research on the feasibility of using AI to detect AI-generated code. My portion of the task is to see the baseline effectiveness of an AI without any fine-tuning. I used a class-given Carleton exclusive free AI model to be able to handle the number of individual queries, and looping it through the 100000 samples in the data took more than a day of continuously running the code, and I luckily have 2 laptops, or I wouldn’t have had any devices to use in my classes. The video below shows how slow the progress is.

[Click to access the presentation Slides of our results]