On the Consciousness of Intelligent Systems
Advances in Artificial Intelligence and Machine Learning 2026 DOI: 10.54364/aaiml.2026.64335 (opens in new tab)
Study at a glance
AI-extracted from the abstract| Characteristics | Review Peer reviewed |
|---|---|
| Key findings | The authors argue that simulated consciousness in artificial systems should be distinguished from natural consciousness, and that interpretable and explainable AI are important for approaching it. They propose fuzzy set-based knowledge representation, including fuzzy concept lattices, ontologies, and description logic, as a way to approach some forms of consciousness and to represent emotions and perceptions. |
Abstract
The consciousness of intelligent systems has been discussed for decades. The ongoing developments of Artificial Intelligence make this concept even more crucial. We review and discuss recent views of the consciousness of artificial systems, mainly based on theories of the consciousness of human beings taking into account phenomenal consciousness, access consciousness and computational functionalism. Distinguishing simulated and natural consciousness, we point out the importance of interpretable and eXplainable Artificial Intelligence to approach the first one. We focus on Large Language Models and Robotics. We also present discussions on the attribution of a moral status to artificial systems. We finally propose to use a fuzzy set-based knowledge representation to approach some forms of consciousness in intelligent systems, following existing directions of research on fuzzy concepts lattices, fuzzy ontologies and fuzzy description logic, as well as the use of fuzzy methods to represent and manage emotions and perceptions, in particular to identify and simulate sentiments in textual or facial expressions.