An algebraic proof can be done starting from the identity $\sum_{i=0}^n\binom{n}{i}^2=\binom{2n}{n}$, which is proved here. Then use the fact that for $0<i<n$ you have $i\binom{n}{i}=n\binom{n-1}{i-1}$. This shows that indeed
$$\sum_{i=0}^ni^2\binom{n}{i}^2=\sum_{i=1}^{n}n^2\binom{n-1}{i-1}^2=n^2\sum_{j=0}^{n-1}\binom{n-1}{j}=n^2\binom{2(n-1)}{n-1}.$$
As for your combinatorial interpretation; the number $i^2\binom{n}{i}^2$ can indeed be interpreted as the number of ways to choose subsets $A,B\subset\{1,\ldots,n\}$ with $|A|=|B|=i$, and then choosing $a\in A$ and $b\in B$. I do not see how this leads to the number
$$n\cdot n\cdot2^{n-1}\cdot 2^{n-1}.$$
Perhaps you did not take into account that $|A|=|B|$?
Of course the number of ways to choose $A,B\subset\{1,\ldots,n\}$ does not depend on the choice of $a$ and $b$, but only on $i$. Indeed the number of ways to choose $A$ and $B$ is $\binom{n-1}{i}$ each, which shows that
$$\sum_{i=0}^ni^2\binom{n}{i}^2=n^2\sum_{i=0}^n\binom{n-1}{i}^2,$$
where $n^2$ is the number of ways to choose $a,b\in\{1,\ldots,n\}$.
This reduces the problem to finding a combinatorial interpretation for the identity
$$\sum_{i=0}^n\binom{n-1}{i}^2=\binom{2(n-1)}{n-1}.$$
Consider choosing $n-1$ balls from a total of $2(n-1)$ balls, of which $n-1$ are blue and $n-1$ are red. There are clearly $\binom{2(n-1)}{n-1}$ ways to choose. We can split up the count by the number $i$ of red balls chosen; then we must choose $n-1-i$ blue balls and hence
$$\binom{2(n-1)}{n-1}=\sum_{i=0}^{n-1}\binom{n-1}{i}\binom{n-1}{n-1-i}=\sum_{i=0}^n\binom{n-1}{i}^2.$$