"CREW ALERTNESS MONITOR The FMC continuously monitors the activation of switches on the mode control panel, EFIS control panel, display select panel, CDUs, and VHF/HF push-to-talk (PTT) switches. When a predefined time elapses after the last control activation, the EICAS advisory message PILOT RESPONSE is generated. The PILOT RESPONSE message can be cleared by pushing any control on any of the monitored systems or panels. If there is still no response after a short time, the EICAS caution message PILOT RESPONSE is displayed. If there is still no response, the warning message PILOT RESPONSE is displayed. Any control activation on the MCP, EFIS control panel, display select panel, CDUs, or VHF/HF transmitters resets the PILOT RESPONSE message. The PILOT RESPONSE message is inhibited at all altitudes below 20,000 feet while the aircraft is in a climb and the flaps are not up."
If the autopilot is detecting the pilot is drowsy I'd rather it let the pilot sleep than prod the pilot awake into further drowsiness.
Crashes are tremendously bad for business. Also, the laws about pilot rest are incredibly strict. That doesn't mean that pilots won't break them, but it does mean that the "economic incentive" aspect is pretty much moot.
1) Flying is incredibly safe
2) Tired, fatigued pilots crash planes
3) Airlines are low-margin businesses
4) There is no economic incentive to have pilots fly less and sleep more
Since there will always be a tension in allowing pilots to sleep more and work less (thus costing airlines more money) we have to conclude that pilots are going to fly without sufficient sleep.
To make the system more resilient to failure, we should take the ad-hoc napping that pilots do now (which is undeniably adding some safety via fatigue reduction) and add a blame-free procedure around it to ensure safety (e.g. when one pilot goes to the bathroom the other one puts on their oxygen mask in case of explosive decompression; something similar should be done for the awake pilot.)