“Author Ken Liu asserts that large language models are demonstrably intelligent, citing their ability to pass the bar exam and achieve perfect scores on the SATs as evidence.”