| Project ID |
BITS-SRIP/6CCCA0/2026 |
| Project Title |
Method and Model Development for Distinguishing Human-Written vs LLM Generated Content |
| Project Description |
The rapid adoption of large language models (LLMs) has made it increasingly difficult to distinguish between human-written and AI-generated content in academic writing, news articles, social media, and online platforms. This raises critical concerns related to academic integrity, misinformation, content authenticity, and trust in digital communication. This project aims to develop robust methods and machine learning models to automatically detect whether a given text is written by a human or generated by an AI system. The project will explore linguistic, stylistic, statistical, and representation-level cues that differentiate human and AI text, and will design hybrid detection models combining classical NLP features with modern deep learning approaches. Objectives are:
Identify linguistic and statistical patterns that distinguish human vs AI text
Build machine learning and deep learning models for detection
Study robustness across domains, topics, and AI models
Analyze failure cases and adversarial scenarios
Explore explainability and confidence calibration of detectors
OUTCOMES:
A curated benchmark dataset of human-written and AI-generated texts across multiple domains (essays, news, answers, reports, etc.).
A set of baseline and advanced detection models
A systematic evaluation of:
Cross-model generalization (GPT, Claude, etc.)
Cross-domain robustness |
| Project Discipline |
Computer Science, AI and Data Science |
| Faculty Name |
vinti agarwal |
| Department |
Department of Computer Science & Information Systems |
