Artificial consciousness remains a nascent field. This paper outlines theoretical assumptions that could help address phenomenal consciousness—the subjective, first-person experience of being. It identifies technological and theoretical obstacles facing researchers and argues that artificial consciousness must confront how phenomenal consciousness arises in a physical world. The authors suggest that externalist models, which locate mental content partly outside the brain, currently offer the most promising path forward.
Pessimism about artificial consciousness (AC) is unfounded, resting on misunderstandings of AI and ignorance of possible roles AI might play in reproducing consciousness. Through conceptual analysis, the paper shows that common objections fail against neglected possibilities for AC: prosthetic, discriminative, practically necessary, and lagom (necessary-but-not-sufficient) AC. Three strands of the author's work—interactive empiricism, synthetic phenomenology, and ontologically conservative heterophenomenology—illustrate these distinctions and defenses.
If consciousness arises from neural electro-chemical interactions, then building a self-aware machine may be possible once the brain's functional behavior is fully understood, mathematically modeled, and artificially implemented. This paper explores technical and philosophical issues surrounding the development of a conscious artificial brain by posing and discussing hazardous questions and answers.
Synthetic methods in science can aim to either instantiate a target phenomenon (strong approach) or simulate key mechanisms underlying it (weak approach). The strong approach assumes a mature theory, while the weak approach helps develop such theories. The authors argue that artificial consciousness is best pursued as a weak means of theory development in consciousness science, not as a strong axiom-driven project to build a conscious artifact. Like other sciences of the artificial, artificial consciousness can elaborate possibilities and limitations of candidate mechanisms, transform properties into mechanism-based criteria, and potentially unify distinct properties via new mechanism-based concepts. The arguments are illustrated by discussing both axiom-driven and neurobiologically grounded approaches to artificial consciousness.