In general, we say a transformation is a symmetry of a theory if it leaves the action invariant, i.e. if
$$S \to S' = S,$$
up to, perhaps, a boundary term (b.t.). However, it is known (see e.g. this post) that the equations of motion of a theory are left invariant if a weaker condition is satisfied. Namely if the action is invariant up to a b.t. and a scaling, i.e. if
$$S \to S' = \alpha S + b.t.$$
This generalized notion of symmetry does not fit well in Noether's theorem. In particular, if I understood correctly, there exists a conserved Noether current only when $\alpha=1$ (see this post). My question is: Why do we care? I mean, symmetry is a fundamental, physical notion, while conserved current is only a mathematical artifact that may facilitate computation. So then, why do we stick to Noether symmetries, which have an associated conserved current? Why don't we seem to care for these generalized symmetries, which leave the equations of motion unchanged? (At least at the classical level, since quantization requires the action to be strictly invariant.)