And this “real world” AI aircraft/AI ATC research is likely coming from some of the Best and Brightest in the industry, (I’m assuming that this means folks like NASA, VPI&SU, Langley, Cal Tech, MIT, Princeton, UCAL Berkeley, DARPA and the rest of the DOD, and so forth), who have essentially bottomless budgets and access to Einstein/Hawkings levels of genius.
And if THESE GUYS are finding it a bit chewy, it makes all the work done by MS/Asobo/Blackshark all the more impressive by comparison. Not that they, (MS, Asobo, Blackshark, etc.), are fly-weights, but they’re not even in the same league as the DOD and top-flight universities with access to the United States Treasury - and possibly other country’s treasuries as well.
Frankly, I find access to the fruits of this kind of research, even if it’s only in a “mere game”, to be quite fascinating.
Before you folks decide to continue bashing MS & Co. for their seeming AI failures, please let me suggest that you get yourself a reasonably competent robot, (The Modular Robotics/Dexter Industries GoPiGo is a great choice - https://gopigo.io, and it’s a lot less expensive than the nearest competition), and then go try some simple AI, like a “simple” autonomous robot that can wander around a room, learn where things are, and find it’s way back to its charging station. Voice commands are optional. Just some “simple” hardware like a LIDAR, maybe an Oak-D, Lite, and some “simple” ROS programming, that’s all. Once you get that done, go on over to the Dexter Industries forums, and post about it there.
Note that I am not asking you to implement it in a globally large solution space (like MSFS is expected to do), just your living room - with a few chairs getting moved around occasionally - and the occasional cat or two - nothing all that complicated, right?
Go ahead. Try some SIMPLE AI for yourself. Publish your results and then come back here to complain about MS’s failures.
OK?