@taylorlorenz I wonder if the speech-to-text errors are biased like LLMs and will shift people’s thinking by changing private communication