Evaluating ChatGPT’s Consciousness and Its Capability to Pass the Turing Test: A Comprehensive Analysis
Matjaž Gams, Sebastjan Kramar
Abstract
This study explores the capabilities of ChatGPT, specifically in relation to consciousness and its performance in the Turing Test. The article begins by examining the diverse perspectives among both the cognitive and AI researchers regarding ChatGPT’s ability to pass the Turing Test. It introduces a hierarchical categorization of the test versions, suggesting that ChatGPT approaches success in the test, albeit primarily with na?ve users. Expert users, conversely, can easily identify its limitations. The paper presents various theories of consciousness, with a particular focus on the Integrated Information Theory proposed by Tononi. This theory serves as the framework for assessing ChatGPT’s level of consciousness. Through an evaluation based on the five axioms and theorems of IIT, the study finds that ChatGPT surpasses previous AI systems in certain aspects; however, ChatGPT significantly falls short of achieving a level of consciousness, particularly when compared to biological sentient beings. The paper concludes by emphasizing the importance of recognizing ChatGPT and similar generative AI models as highly advanced and intelligent tools, yet distinctly lacking the consciousness attributes found in advanced living organisms.