From what I’ve read, the CPU has to send the data to the GPU before the GPU can do anything with it. But, if that’s the case, won’t any time saved using the GPU be negated by the time taken to transfer that data? Wouldn’t that kind of defeat the whole purpose of a GPU?
It depends. If it's a lot of data and a little bit of computation, yes, there is no net savings. If it's a little bit of data and a lot of computation, no, there is a net savings. It all depends. Normally a GPU is used for situations where the amount of computation time saved greatly outweighs the time to transfer the data -- such as training machine learning models, or rendering complex graphics.