Old connections and alumni. This is interesting about code complexity. But as the article says, consider the source, a source management company. I remember the only software we every shipped that was actually 100% bug free was done by one person, an architect Dev and tested by me the manager AND it was very small. Even the limits and stress testing was rock solid, and it localized in to 27 languages. https://arstechnica.com/gadgets/202...ing-100x-more-code-now-than-they-did-in-2010/
Ever wonder about your car? I have a Honda civic hybrid that has a bug. Under 35MPH, it will occasionally shut down the power steering. Restarting resets it. At one point the windshield wipers were haywire too, stopping mid windshield. Not sure if the two are related, but they were co-incident a few times. Honda never could find the issue.
Speaking of AI & machines flying planes, watch the video: Good luck AI powered planes doing something like this. PS: Some might argue that AI would be much better than humans in such critical & difficult situations, but I don't think so, because in uncertain situations like this, there are so many unknowns to deal with in a very short period of time which makes human experience invaluable to AI or ML.
So, You don't think current AI/ML technology would be able to correct for wind and the angle of the plane and wheels in relation to the runway? You don't think that they program the ML with data from actual flights ... like this one. Data that includes the pilots' maneuvers? You don't think that a computer can calculate the physics involved in flying, faster, and more accurately than a human?
I do believe that computers/machines can process lots of data much faster than humans & that they can calculate physics/mechanics much faster & more accurate than humans. I also do believe that ML gets data from actual flights as well as simulations, tons of them actually in order to gain the human experience as much as possible and be prepared for any scenario they might face. BUT: I believe that the human experience is invaluable and will never be matched by machines, no matter how much data you teach them through ML. Chess is a 64 squares board with 32 pieces in total. It is estimated that for a 40 moves games, 10^120 games are possible. After 10 moves only, there is 70 trillion possible combinations after that & that increases at a staggering rate with every move beyond that. And this is just a small chess board with 2 players and a bunch of pieces. It took computers a huge amount of time (Even with ML and AI) to rival & beat experienced professional players in chess and even up until this day, the mastery of the game has not fully moved from humans to computers. In doing more complex things, like flying a plane or driving a car, one might think that the possibilities are limited, but the opposite is true. The possible combinations/variations in any given situation are huge with so many variables, both internal & external. Now, you might think that it's the speed & accuracy of calculations which can only resolve such critical situations, hence computers win such a game for sure, but it's not. Acting quickly is one thing, doing calculations quickly is another thing, but searching for the optimal solution to this particular situation & implementing it is the ultimate objective. A computer will look for the optimal solution & find it quickly, then will put it to execution. Then inputs & variables will change, triggering re-evaluation of the situation, then maybe another solution comes up, making the solution which is currently in execution not viable, etc...How can a machine deal with that? Human experience works differently, it's not always pure science, physics & math as computers do it. Part of it is art (If you may call it). A human can choose a solution which seems impossible when evaluated by a computer because of laws of physics or mathematical calculations, yet, when such a solution gets implemented by a human, it works!!! How many times have you seen car racing drivers push their cars beyond the limit where the manufacturers of those cars would never believe it can be done without breaking the car?? Do you know how many impossible (From a physics/biology point of view) combat flight maneuvers happened in the last 100 years? I understand, there is a fine difference between doing crazy stuff and very difficult stuff that seems impossible, so I am not talking about crazy stuff, I am talking about scientifically borderline or nearly impossible. Sorry for the long post.
So, other than that, what other "things" are required to land a plane in the wind? ...That a computer can't do as well, and likely better, than a human? I don't think the goal of the data is the "human experience." The data is for calculations. Calculations that go beyond the capabilities of the human experience. A calculator doesn't try to be human. It has capabilities beyond the human experience. It wants to be the best calculator it can be. We build planes not to have the bird experience. The goal of planes extends far beyond the inspiration of the plane. Far beyond what we first studied in order to learn how to fly a plane. Birds can't compete with planes. And humans can't compete with machines ... with regard to specifics tasks. Again, I'm not talking about the human experience, and I'm not sure exactly what you mean by that. The point is: landing a plane in the wind is a physics problem; quickly and easily solved faster by machine, than any mortal could. Again, if you are suggesting that there is some mystery element needed to land a plane in the wind; an element that only exists in humans; and can't be replicated and improved upon by machines; please explain. Right now, from where I sit, landing a plane in the wind is still a physics problem. "Huge" is relative. Regardless, that has nothing to do with my assertion: A machine can land a plane in the wind better than humans. There are more possibilities in a game of GO. And there are more possibilities while driving in a populated city vs. landing a plane in the wind. Machines are better at finding optimal solutions. How fast can you find the optimal solutions to this: 54+90-65/3+8!/2^(6.3)=15.6x^3 How fast can a machine find the optimal solution? Besides, most real world problems don't require the optimal solution, only a 'good enough' solution. How can a machine deal with that? Easily. Just recalculate. Just like your GPS navigation system does when you miss a turn. Just like a calculator does when you give it a new problem. Just like a chess game does when you make a move it didn't expect. Just like the computer opponents do on video games. Etc. Etc. Did you solve that math problem in your head yet? If I change a number in it, how fast would you be able to recalculate? My premise is about a machine landing a plane in the wind. As far as I know, that deals with the laws of motion, etc. As far as I know, it has nothing to do with the human experience. Now, if you had posted a video explaining how paintings require the human experience, and that no machine could paint as well as a human; I wouldn't have responded. Landing a plane in the wind is about physics. Never. But I bet the engineers knew the car wouldn't break. Whether the car breaks is physics. A human can only accelerate, brake, and steer (and shift gears etc.) These things are all physics. The metals used in the vehicle have known properties. How much stress they can take is known, and can be calculated. Nothing is voodoo about it. None. Also, my premise is regarding landing a plane in the wind. I'm talking about landing a plane in the wind. You moved the goal post on me. Happens all the time on message boards.
The computer and the human can both try and land in the wind. But the human pilot should have known to not bother attempting to land with such a strong crosswind component. The risk of something going wrong (like a sudden drop in wind speed which would have caused the plane to crash into the ground), is too great. So both of them lose by even trying it in the first place.