It might be helpful to think about this situation this way:
What is your evidence that oxygen exists?
How would you test that?
What tools would you need to test that?
How do you know those tools work the way that you expect?
What is your evidence that oxygen exists?
How would you test that?
What tools would you need to test that?
How do you know those tools work the way that you expect?
But you'd have to really care about the quality of your model to do all that work.
Models derived from authority may lack such grounding but they are way more convenient.
And convenience is king.