There could be a number of things that can make that sound.
One way is to upload the raw vocals into a pitch & time editor, and there quantize the vocals (but not all the way) and do further manual editing, making them sound somewhere between
natural singing and autotuned.
Another way is to copy the vocal track and apply complete autotuning, use a stereodelay with a longer delay to prevent phasing in mono, and blend it in with
the original. You tend to keep the autotuned channel not too loud, or it will sound unbalanced or loose by the dim stereodelay.
Or if you have several takes, you can in theory process 2 different takes this way and pan them hard left/right and blend them in, probably making them much more balanced and tight.
Though since they are completely autotuned they can get too similar to eachother, and by that maybe turn into mono sometimes. I haven't tried this, I'm just speculating.
Lastly they probably go for the method where they record several takes and then choose the best parts of each to create a "megatake", and that can perhaps make it sound unnatural sometimes.
Then of course, the right decisions of mixing techniques can enhance this effect.