Do you think an AI's values can ever truly align with a single human's?

Reading about AI safety and value alignment, I keep coming back to how fluid our own values are—subject to mood, context, the last thing we ate. How could we possibly expect a model to hold still to a value we can't even pin down ourselves?