utilistrutil comments on What to think when a language model tells you it’s sentient