3D scanners are getting better, and more importantly the software behind them is getting better. Right now a lot of the geometry from scanning is pretty ugly, but there is a lot of talk about improving it.
The latest version of Meshfusion for Modo is an absolute marvel in how it can do really complicated mesh booleans involving several pieces of organic looking meshes and then somehow, with some kind of voodoo black magic, produce clean sub-division ready geometry. It even puts in perfect edge loops and chamfers the edges etc. ZRemesher (the retopology tool in ZBrush) is pretty impressive too… we usually end up fixing some issues with the low-poly mesh in Maya, but overall, I’d say it produces pretty good geometry. How hard would it be to import a scanned mesh into ZBrush, touch up the bad areas, then use ZRemesher to make a game-ready mesh? I think it would be pretty doable.
I’m not arguing that 3d scanning is replacing anyone’s job yet, but I wouldn’t be too quick to say that 3d scanning won’t become a major tool for game makers in the future. I’d say maybe 5-10 years from now.