I was working on this problem from my Complex Analysis book:
First, I assume that $a, b, c$ are not literarly elements of $\mathbb{R}$; we know that $\mathbb{R} \not \subset \mathbb{C}$ and my book hasn't defined multiplication between elements of $\mathbb{R}$ and $\mathbb{C}$, so I'll assume it meant $a, b, c \in \mathbb{C}$ with Im $a, b, c = 0$.
Now if Re$(b^2-4ac, 0)$ is a negative real number, then we can check that $(0, 1) \cdot (\sqrt{-b^2+4ac}, 0)$ is a solution to $z^2 = (b^2-4ac, 0)$. So why do we need to define $\sqrt {b^2-4ac} = i \sqrt {-b^2+4ac}$ if $b^2-4ac < 0$, when in fact it couldn't even be anything else, if it was to remain consistent with the rules of multiplication already given?