Why do we need a GPU?
Imagine coloring a thousand little boxes. One capable helper can do it, but a team can work on many boxes together. That is the basic idea behind a GPU.
Two processors working together
Runs the program and coordinates its tasks
Handles many suitable pieces of work together
Why we added GPUs
A screen contains tiny dots called pixels. Drawing a changing picture requires lots of similar calculations about their colors. GPUs were built for that work: graphics processing unit is what the name means.
AI also involves large amounts of suitable arithmetic. The same ability to handle many calculations together makes GPUs useful there.
We still need the CPU
The CPU handles a wide variety of tasks and decisions. The GPU helps with work that can be divided into many similar pieces.
Real CPUs also work in parallel. The helper analogy explains the emphasis, not an absolute rule about what either processor can do.
Try it yourself
Would you replace the CPU with a GPU to run the whole computer?
Check your answer
No. They work together. Adding a specialist does not remove the need for a general-purpose processor.