The normality assumption is critical in statistics for parametric hypothesis testing of the mean, such as the t-test. As a result, we may believe that these tests are invalid when the population is not normally distributed. But, if our sample size is large enough, the central limit theorem comes into effect and creates sampling distributions that are close to normal. Because of this, we can apply parametric hypothesis tests even if population is not normally distributed, as long as sample size is large enough.
Then what is the use of non-parametric tests?