If your friend knows that you study Bayesian reasoning, why would he expect point 1 to be news to you? Bayesian statistics, etc. deserve conscious study in part because they require conscious study. If we already reasoned perfectly there'd be no rational reason to try and improve it.

Point 2 is more interesting, but only because it superficially resembles a much more serious concern. If psychologists are sure that our reasoning "is mostly unconscious", that's still not the same as saying that it "must always be mostly unconscious despite conscious attempts at intervention". The former proposition just means that part of instrumental rationality needs to be learning how to identify the most critical decisions to treat with less intuitive and more formal reasoning. The latter proposition would mean that studying rationality is pointless to begin with, but it would also mean that studying engineering is no better than "guesstimate the cathedral designs and keep the results that don't collapse". Clearly we're capable of improving over unstudied human reasoning in some limited forms of decision making, and it makes sense to try and push those limits out as far as we can.

And while it may sound off the wall to someone like your friend outside the context of this site: even if some techniques for rationally improving decision-making are too hard for an un-assisted human brain to implement, that's just similar to the fact that many techniques for modern industry are too hard for un-assisted human muscle. That just makes them more worth study, as a necessary prerequisite for figuring out how to design the assistance.