The Klein $4$-group is generated by $2$ elements $a$ and $b$ satisfying the relations you put on them when defining $G$, but that does not mean the group $G$ can be the Klein $4$-group. After all, the trivial group also has $2$ generators $a$ and $b$ satisfying the relations you put on them when defining $G$ but you would not have argued that $G$ is the trivial group, right?
The fact that multiple nonisomorphic groups can have generators satisfying the same relations in a group presentation is not an inconsistency: all those groups will be quotients of the biggest such group, by which I mean the group actually defined by the group presentation. By definition, a group presentation is a group of the form $F/N$ where $F$ is a free group with generators corresponding to each of the generating elements in the group presentation and $N$ is the smallest normal subgroup of $F$ containing the relations.
Example. The dihedral group $D_n$ of order $2n$ can be described as $\langle x,y|x^n = 1, y^2 = 1, yxy^{-1} = x^{-1}\rangle$, which is $F_2/N$ where $F_2$ is the free group on two elements $x$ and $y$ and $N$ is the normal subgroup of $F_2$ generated by $\{x^n, y^2, yxy^{-1}x\}$. Every quotient group of $D_n$ also has two generators satisfying the same relations as in the definition of $D_n$, but proper quotient groups of $D_n$ will have "more relations".
It is not always easy to tell how big a group is when it is described by a group presentation: some group presentations can be the trivial group and that can be very tricky to show. See here.