When studies are run, we aim at estimating values that are true for the population. However, we often cannot record data from everyone in the population, which is why we rely on drawing a random sample from the population. For example, while we may want to estimate the average difference in height between all men and all women in the world, we cannot possibly measure the height of all men and women in the world. Therefore, we draw a random sample of men and women. Let's say we collect data from 100 men and 100 women. The study reveals the average difference in height we find in this sample of 200 people, but it does not tell us what the true difference in height in the population of all men and women in the world is.
If we drew random samples of 200 people from the population of all men and women in the world again and again and again, and assessed their average difference in height each time, we would find a range of values. This range of values represents our estimates for the height difference in the population of all men and women in the world.
We refer to this range of values (interval) as the confidence interval. We want to make sure that it includes the true value of the variable we are estimating for the population sufficiently often. If we refer to a 95% Confidence Interval ('CI'), this means that our range of estimates from random samples contains the true value of the population in 95% of all cases.
If we calculate a CI from one study that we have run, it tells us the probability (e.g., 95%) that the CIs of repeated future samples would contain the true population value.