
The strange link between AI hallucination and creativity
The story in brief
In his 1989 book *The Emperor’s New Mind*, Nobel laureate Roger Penrose told a story about a supercomputer switched on for the first time, where a boy asked a question. Penrose argued that imagination depends on venturing beyond known information—a concept modern researchers echo, noting LLMs don’t function like databases, and hallucination reflects computational limits between accuracy and surprise.