Here is how I believe the test went-

Wing was set to release. Therefore the software already knew what type of parameters to fly most likely from windtunnel and computer simulations on the exact length of wing etc. Great for a promotional investor type video but little fact for real world situations.

Now I'm not saying that down the road this technology won't be useful....It appears to have great uses......but if I was the pilot.....I know the ejection seat works. Its proven. Would I trust the computer to land the plane with battle damage that most likely won't be such a pre-determined clean break? And how do I know the computers program wasn't damaged by the battle damage?

For me this would be the last option......you know when the checklist states-"there are no provisions for manual bailout if the ejection seat fails". Then I would trust this system.

Once you get below 2k ft your chances of surviving the ejection become less.