Applying Data Sets is a way to make all jobs in a project point to the same databases so the wrong database is not updated. If you update a production database is updating during a test run then this could be embarassing. It could waste a lot of time trying to undo it. Applying Data Sets will only update the default values in a job. It will not change prompts or insert new parameters. Data Sets are grouped into three categories DEV, TEST and PROD by a field called PROJECT_TYPE. You can only apply data sets to projects where the project type matches. This will save you from applying data sets improperly.
The way to apply a data set is similar to copy and paste parameters. The first thing you need to is Select a Data Set to Apply by right clicking in the tree control on the Data Set name you want to apply. Next you find the project or job you want to update and then right click and Apply Data Set. Very easy.
More Detail