Analyzing Human and ChatGPT Responses: A Comparative Study of Transformer Models in Natural Language Processing

Abstract

With the latest developments in artificial intelligence, language models have reached the level that can produce human-like responses. One of the most successful examples of these technologies is the GPT-3 language model developed by OpenAI. This study aims to investigate how similar human responses and artificial intelligence responses are, and which transformer model performs best. In the study, a series of questions were answered using different transformer-based models and compared with the answers given by humans. The highest performance was obtained with the roberta-base-openai-detector model. This study provides researchers with a different perspective to understand the difference between artificial intelligence technologies and human responses. © 2023 IEEE.