Often autonomous driving system is often seen in misused or fun experimentation. However the driver-automation interaction possibilities are often neglected. The Study on  Super Cruise by MIT reveals both the potential and the drawbacks of self-driving systems.

Hands off the wheel
In the world of autopilot mode, the driver-automation interaction is such that it is usually preferred on highways or lane-marked roads. As the technology is designed in such a way, to identify the lane marks and drive accordingly on the road. It was expected that the drivers also utilize the autopilot mode similarly. However, it is observed that the engagement and disengagement of autopilot mode are patterned differently.
Furthermore, the factors to consider for using the autopilot are location, the purpose of driving, and comfort. Depending on what the situation is, preference can vary because the autopilot mode has limited speed goals. While not everyone takes the time to plan according to the autopilot more. Here the driver-automation interaction is more need base.
On the other hand, if the driver realizes that they are sleepy or not in a stage to be fully attentive. Then they could possibly take autopilot assistance. Which indeed makes sense to use, avoid accidents and also reach the destination.
The methods used
Fourteen drivers from the Boston area of Massachusetts participated in the study. They didn’t have any bad records with respect to accidents or any such activities, and the average age of the group was 42.
The study was for MIT Advanced Vehicle Technology data collection. When the data collection started in 2020, they preferred to use the Cadillac CT6′s Super Cruise. It was on level 2 automation driving systems, much advanced than its then competitors. And the data was taken on a monthly basis as the cars were designed in such a way to collect specific details. It included when the driver was on automation when they weren’t and so on.
Determining the driver-automation interaction solely with this study is not a solution. However, the results gave a new perspective.
While all kinds of minor details like speed and distance were considered. Their result is focused on how the drivers preferred to use the autopilot. It is possible that the drivers analyze their results using autopilot. While disengaging the Level 2 autopilot mode, they usually seem to shift to Level 1. Rather than fully going to manual mode, level 1 was preferred 73% of the time.

Further, the shifting of usage also depending on the lane markings and the ability to change for the autopilot mode. It is highly possible that the drivers too are testing the potential of advanced driving. Which indeed plays a major role in driver-automation interaction.