Having recently experienced talking to AI voice bots (for customer support, not for interviewing), it's bizzare that you don't know what they know and they're just making up.
If you ask "Will the role expect me to XYZ" the bot probably only has limited context from a job posting 1 pager, so you can't actually trust it or try to align with it's goals/experiences.
What I can't trust it to do is properly represent me as a candidate the same way a human recruiter can.
I don't expect answers to random or ad-hoc questions. The A person can clarify a question that is part of a candidate screen. The AI "interviewers" so far have not.