In my version of reality, the science must account for how pundit brains compute as well as scientist brains compute. You neglect the fact that in math (or logic) the assumptions are not tested for truth. They are by definition assumed true. See Descartes. In computation explanations you can have completeness. It is a positive feature of deep learning explanations that they make mistakes like real people do, but the mistakes are observationally like those that people make, as pundits or scientists, both, not one or the other, but clear on why and how. One of my references perfectly defines what I mean by direct observation, without assumption.

I was debating, but decided not to put into my comment, a comparison with other human roles like 'engineer,' 'theologian,' 'story teller,' 'fiction writer,' 'great orator' etc., as covered in my book on "How to get your (good) lies back." Perhaps I should have. Yes, science is a social process, sadly.

https://medium.com/liecatcher/physicist-admits-they-all-lie-alot-d492adc65fe2

Truthcourt.net provides the best solution possible with human brains. That too may elude your irrational bias. I point to Darwin and other scientists who did not use math as the basis for scientific explanation.