Turning to the use of AI in legal research, Kant warned that a fluent and convincing answer was not necessarily an accurate one