I'd say it depends on the standard for knowledge. If we mix up mathematical and empirical standards, we wouldn't "know" anything without proof that licenses certainty. So knowledge would deal only with necessary rather than probable propositions. But that's a Cartesian confusion. Elsewhere, I argue against agnosticism on similar grounds, by the way.

https://medium.com/interfaith-now/thomas-huxleys-scientism-and-the-agnostic-s-false-modesty-20862b2be550?sk=96ffb743628853c05ab4cb31bee30534