I don’t think the microphone quality will be an issue. Drive through intercom systems have been around a while.
Accents might be, but that was an issue with things like Siri since the beginning and I can only imagine the technology has gotten better.
I would expect the AI to integrate with a real person in case there are issues with the AI understanding the order.
Unrelated to fast food, I think this technology is going to be implemented in a lot of places. An insurance company receiving an email with a description about a claim, for instance, will automatically put the claim in the system with all the details from the email. A user will do a follow up for accuracy, but as time goes on and that gets more dependable, the user will be much less necessary for this…