I usually find the AI ethicists and alarmists tend to oversimplify things. This @emilymbender response to a NYT piece I thought did a good job capturing some nuances surrounding large language models, at least good as a snapshot of shortcomings at a moment in time.