This is easy yet necessary. In particular the specific wording of the alternative hypothesis tells us whether we need to perform a one-tailed or two-tailed test.
It is essential that the level of significance is decided before the data is collected. If we decide the significance level after the data is collected, we may pick a stronger/weaker level based on the data in order to prove our point.
The critical region tells us what values of the test statistic will result in the null hypothesis being rejected hence the alternative hypothesis being accepted and depends on the chosen significance level.
It is through calculating the test statistic and seeing whether or not it is in the critical region that lets us know which hypothesis to accept.
If the test statistic is in the confidence interval then accept $H_0$. If the test statistic is in a critical region then accept $H_1$.
Null and Alternative Hypotheses