Enclose your integration in $\displaystyle{\left[-1,1\right)^{\,2}}$. The Monte-Carlo integration becomes $\left(~overline\ \overline{\phantom{AAA}}\mbox{means average with an uniform distibution over}\ \left[-1,1\right)^{\,2}~\right)$
\begin{align}
S_{N} & = \sum_{i = 1}^{N}x_{i}^{2}
\left[\vphantom{\Large A}\left\vert x_{i}\right\vert +
\left\vert y_{i}\right\vert \leq 1\right]
\\[2mm]
\overline{S_{N}} & = N\ \overline{x^{2}
\left[\vphantom{\Large A}\left\vert x\right\vert +
\left\vert y\right\vert \leq 1\right]} =
N\int_{-1}^{1}\int_{-1}^{1}{1 \over 4}
\left[\vphantom{\Large A}\left\vert x\right\vert +
\left\vert y\right\vert \leq 1\right]x^{2}
\,\mathrm{d}x\,\mathrm{d}y
\\[5mm]
& \implies \int_{-1}^{1}\int_{-1}^{1}
\left[\vphantom{\Large A}\left\vert x\right\vert +
\left\vert y\right\vert \leq 1\right]x^{2}
\,\mathrm{d}x\,\mathrm{d}y = 4\,{\overline{S_{N}} \over N}
\approx \bbox[10px,#ffd,border:1px groove navy]
{4\,{S_{N} \over N}}
\end{align}
The following code is a $\texttt{javascript}$ script which can be run in a terminal with $\texttt{node.js}$:
"use strict";
const ITERATIONS = 10000;
let i = 0, theSum = 0, x = null, y = null;
while (i < ITERATIONS) {
x = 2.0*Math.random() - 1.0;
y = 2.0*Math.random() - 1.0;
if (Math.abs(x) + Math.abs(y) <= 1.0) theSum += x*x;
++i;
}
console.log(4.0*(theSum/ITERATIONS));
A typical "run" yields $\bbox[10px,#ffd,border:1px groove navy]{\displaystyle 0.3302123390009306}$. The exact result is
$\bbox[10px,#ffd,border:1px groove navy]
{\displaystyle{1 \over 3}}$.