Artificial Intelligence and Consciousness: Limits and Modern Perspectives

Consciousness is defined as subjective experience or the capacity for self-reflection, while intelligence is the ability to learn, solve problems, and adapt. Weak AI, including generative models like GPT-5, Gemini 2.5, and DeepSeek-V3.2, demonstrates high-precision pattern prediction and linguistic ability but lacks genuine understanding—a feature of strong AI. The paper examines whether computational processes can constitute real thinking, referencing Gödel's incompleteness theorems, Searle's Chinese Room argument, and the Turing Test. Analysis of contemporary language models' responses to Gödelian questions and reasoning tasks shows that despite progress, the question of machine consciousness remains unresolved and is a subject of philosophical debate.

Study at a glance

Abstract

Summary This paper provides a review of selected concepts concerning consciousness, intelligence and artificial intelligence, focusing on their interrelations and interpretative limitations. The aim of the paper is to organize key definitions and viewpoints, and to highlight central issues related to the question of whether conscious machines can ever emerge. Consciousness is often defined as subjective experience, as the capacity for reflection on one’s own mental states, or as an emergent property of complex biological systems. Intelligence, on the other hand, is interpreted as the ability to learn, solve problems, adapt to changing conditions, and control cognitive processes. The development of computational technologies has given rise to weak artificial intelligence, encompassing algorithmic and machine learning systems that can model and predict patterns with high precision. Within this category, generative artificial intelligence, represented by large language models, demonstrates impressive linguistic capabilities but lacks genuine understanding – a feature associated with strong AI. The paper discusses whether computational processes can be equated with real thinking, referring to Gödel’s incompleteness theorems, Searle’s Chinese Room argument, as well as the Turing Test. This review contributes by integrating classical philosophical arguments with a comparative evaluation of contemporary language models (GPT-5, Gemini 2.5, DeepSeek-V3.2), examining their responses to Gödelian questions and reasoning tasks. The analysis indicates that, despite significant progress in building artificial intelligence systems, the question of their potential consciousness remains unresolved and continues to be a subject of profound philosophical debate.