I'm not a tech head by any means but from what I understand it means the GPU which is mainly used for rendering graphics can be used for tasks that the CPU normally takes care off, stuff like physics and collison detection.
CPU's are usually large cores which are very fast while GPU's are lots of tiny little cores which aren't as fast. CPU's do tasks one after the other while GPU's do lots of tasks in parallel. By offloading tasks to the GPU you can have these tiny cores perform smaller CPU like calculations along with the CPU. Since they both perform calculations in different ways you can sort of mix and match and code to each one's advantages which will mean better physics on objects in the game i.e. better destruction of buildings, clothe animation car handling etc.
They showed it in the PS4 reveal when they had the GPU run that physics demo with the blue balls dropping from the sky.
Also the fact that the GPU and CPU are on the same chip and share the same pool of RAM wil probably mean there is a lot they can get out of this setup.
Thats where my knowledge ends
Another reason to be excited about next gen though because it opens up a lot of options for the developers.