Unpopular mitigation: everywhere someone would use a macro to abstract over syntax, it should be obviated by introducing or generalising language features to let them abstract over semantics instead (e.g. higher-kinded types, higher-rank polymorphism, first-class patterns…)
We are still a ways off being able to truly do type theory in type theory, so you still need some level of support for automation - be it macro expasion or at elaboration time...