3 ms·

Autopilot can't deal with unexpected situations or instrumentation failures particularly well. Take the Sully landing for example - without specifically setting

• nexuist 7y ago
  Well, it is called the miracle on the Hudson for a reason. Software is only as capable as the hardware it is running on. If you pulled the brakes off a Model 3 and brake-checked it, you'd be hard pressed to find a human or Autopilot version that could solve that problem. In special cases, you might have a lane open to swerve into - like Sully did with the Hudson - but such maneuvers are rarely available. To be fair, dual engine loss on takeoff is also not really a common occurrence. Truthfully, Sully's incident represents the peak of human ingenuity, and I concede that it will take us decades if not centuries to reach a level of automation that can match that. But an autopilot that can react appropriately to 99.999..% of cases? I think it could already exist. Certainly, if dual engine loss is on roadmap, proper software could be developed to ensure the best possible scenario. I think that it's never been a consideration for the industry because these incidents are so rare, and also because that effort is better spent in trying to build more reliable engines. An ounce of prevention is worth a pound of cure and all that.
  • sansnomme 7y ago
    If AI can be cheaper than human pilots, then the cost savings can easily be thrown into massive parachutes for cabins like those in smaller private planes. That, or solid state emergency landing thrusters.