Faulty Artificial Intelligence, or the Sleep of Reason

This perspective argues that the term “AI hallucination” is a misnomer, contending that large language model errors more closely resemble confabulation — active, fabricated outputs — rather than hallucination, which implies conscious perception.