It appears now that the plane that went down over Buffalo was covered in ice and on autopilot. The FAA warned against this in 1994, and it was since airline policy not to do so.
A bit of research quickly finds the Smart Icing Systems Project from the
Aerodynamics and Flight Mechanics Group at UIUC, which has been studying this problem since 1996, and in 2003 developed a set of algorithms to allow autopilot systems to handle icing conditions.
So, that leads then to the following lines of questioning:
when was that airplane’s autopilot system last updated?
is it updatable?
does the airline do updates, if it is?
has the manufacturer implemented ice-condition algorithms?
how is icing reliably detected?
how can I know, as a customer of an airline, if the plane I’m going to be on has the best possible autopilot? (without market demand the tendency is towards cost minimization)
should incorporating best practices into air flight software be mandated by certification bodies or governments?
In general, automatic systems do better than human pilots. The navy has automated landing systems for aircraft carriers that can hit the cable on the deck under full steam in high seas. They do better than the Navy’s human pilots do, so I think I’d rather have that system landing me at O’Hare than a human. Collisions often happen when pilots ignore warning systems. The latest fighter jets can’t even fly without computer control. However, problems like the one in Buffalo will only tarnish public perception of automated aircraft systems.
It appears this incident actually proves that we can’t rely on human judgment in dangerous conditions, but I expect it’ll be spun the opposite way.