Consciousness and Rationality: The Lesson from Artificial Intelligence
Journal of Consciousness Studies June 1, 2022 Philip Woodward
The author reviews three longstanding objections to artificial intelligence: that machines lack consciousness, cannot solve fully general problems, and operate without genuine semantic understanding. The argument connects these problems through the concept of non-syntactic inferences—inferences based on insight into rational relationships among thought-contents. Such insight, the author contends, is only possible through phenomenal consciousness, and it alone gives epistemic justification to those inferences. Until artificial systems can execute inferences guided by conscious states, they cannot be considered intelligent in a literal sense.