Rethinking AI Consciousness: A Logical Argument from Functional Equivalence and Sentience — LessWrong