AI Vs Human–Academic Essay Authenticity Challenge

The appearance of large language models (LLM) including GPT 3.5 and GPT-4 have called into question the integrity of academic writing, as it is becoming increasingly more difficult to distinguish between essays written by humans and those that are AIgenerated. This project seeks to create an answer to this challenge by designing a binary classification model, to identify if an essay was written by a human, or a machine. The dataset contains essays written by humans, which were gathered from the ETS Corpus of Non-Native Written English, alongside essays written by AI from seven LLMs: GPT-3.5-Turbo, GPT- 4o, Gemini-1.5, Llama-3.1 (8B), Phi-3.5-mini, Claude-3.5. Utilizing natural language processing (NLP) techniques and machine learning algorithms, this system will skim for linguistic patterns in academic essays written in English and Arabic in order to separate either human essays from AI induced writing. It is a solution that will support academic authenticity and lessen academic dishonesty with the misuse of AI tools in the university classroom to help our institutions support fair academic practices.