Simple depth based screen space ambient occlusion is just a post process filter doable to any image if you have access to the depth buffer. However for best results you also need normal vector buffer and you only want to reduce the ambient lighting amount, not darken all pixels like the NVidia approach does. Developer implemented SSAO filters will always look better than driver forced ones.
For future cards (think GT300) the ability to force RGSS would be nice. Especially for the myriad PC games these days that lack proper MSAA support.
This approach has several problems. The driver just cannot force the game to render with a larger supersampled resolution. Many pixel precise texture lookups (for example half pixel optimized blur kernels, etc) would completely break if the resolution or sampling coordinates are forcefully altered by the driver. Everything using VPOS would break also. The driver has no foolproof way of detecting which buffers to create/render in supersampled resolution, and when to do the supersample downsample/resolve for each of these buffers. Also simple 2x2 larger resolution rendering and downscaling image before displaying would increase the processing power need for pixel processing by 4x. It's only practical solution for old games (as very few new games run at 240 fps on the resolution majority of the players are confortable with).
