okay here is the right answer:
it is not impossible and not incredible expensive to install an AI driver in
trains.in truth it is even easier to build one for trains then for a car.
This comes from the fact, that the interaction between human drivers and AI drivers is far more limited for trains (among other reasons). Similar things can be said about ships and airplanes by the way.
So why does it take so long, or perhaps will never happen in the future?
...
Money! just think about it. There are over 43 million cars in my country. But the number of trains should be in range of thousands. Development an AI will cost quite a lot. and there is no guarantee that it will be allowed in the near future. Sure they are most likely safer than the human driver. But one dumb error( even it is much more unlikely for the AI) like a dead child can result in temporary ban of it.
and then what? quickly employ thousands of people with the necessary skillset? impossible.
So what about the potential benefit?
The AI will replace thousands of people, this is of course a money saver.
But the AI is no static wonder machine. You still need to employ a huge team to monitor everything, record new data footage, and retrain your agents. The trains+tracks will also get more expensive with the additional sensors and actors.
in the end it will still save the companies some money. But the risk still outweighs the potential benefit.
Bonus reason: The first company to fire the majority or all train drivers will face the negative PR from they customers.
