You can track where the eyes are looking (i.e. the intersection of the eye gaze direction with the surface of the screen), but this typically involves a lot of calibration, and often more than one camera, just to make it reliable. Even if it's reliable, it's horrible to use - they have these on display at conventions all the time, and they're generally used for checking if advertising works.
The other is to track the position of eyes in an image and a rough direction of where they're looking. You can then use this information to augment the image captured from the camera to make it look like the eyes are looking at the camera (when in fact they were looking at the screen, as in video-conferencing). I suspect this would be the more likely technology to make it into a tablet/laptop.