There are a lot of things intersecting here, combination of what technology can/can't do and also the impact that technology has, along with (not surprisingly) people's fears for the future.
Technologically, automated control is nothing new, a lot of the modern era metros (Bart, Washington DC originally) as far as I know rely on automated train control, as for example do the monorails that run at airports and such. Some of them,like the BART system, have an operator but they are a backup in case something goes wrong. The DC system points out some of the issues of technology, they suspended automated operation in 2009 because of an accident with fatalities (the actual automated train control was working, the problem was the sensors they use to detect trains trackside failed, and the train that read ended the stationary train basically couldn't 'see it'...on the other hand, as we have seen with engineer controlled trains, failures of standard track signals have caused collisions, too, whether a human engineer or an automated train control, they rely on feedback systems that may malfunction (everything from fog obscuring the track ahead, to signals not displaying properly).
Most people don't even realize the amount that supposedly "manually" driven things are not. Modern fighter aircraft, everything beyond the early 1970s, are aerodynamically unstable, so the pilot is basically flying a plane where what they request the plane to do is controlled by automated systems, they are not directly controlling the flaps and aerilons or the rudder, rather they request what they want to do and the computer system translate that into instructions to the flight surfaces that are rapidly adjusting them, if you tried to fly one of those 'by wire', literally the pilot tried to control the angle on the elevator or the aerilons, it would crash. Drive by wire exists on vehicles today, where the operator turns the wheel, and it is translated into electro-mechanical control of the wheels by computer system. Modern passenger airliners are much the same, Captain Sully's skill was very real, but when he was using the control yoke on that aircraft, trying to land it, computer systems were handling the actual control surfaces the same way they do on jet fighters from what i know (I am sure of it with Jet fighters, my father worked on the F15 as an engineer, and uncle was an engineer at Grumman who produced the F14, pretty sure with modern era Boeing and Airbus aircraft).
The real problem is what kind of risk are we willing to take and how do we ameliorate it? For example, with the Metro crash, the problem (to me) was a single point of failure, with the trackside sensors there was no way to know if one was defective, and they apparently don't do redundancy (want to know why fighter planes aren't failing all the time despite being controlled by computer fly by wire systems? Because the military requires several levels of redundancy, and the pilot has all kinds of diagnostics that tell them when something is going wrong). A PTC system is only as good as the components on it, if a single bad sensor can cause a crash it is a crap design, or one that cannot tell someone it is failing (for example, with trackside block sensors, if they had half a brain they would have them networked, and have some sort of "challenge/response", and if a sensor doesn't ack the message, it is reported as down and appropriate action taken, like warning trains in that block that the system is down and to use different rules . If a train is automated control, this is likely where an engineer would take over, likely the automated system would stop the train if told there was a problem, and the engineer would take over under some sort of other rules I would assume). My concern about automated train control and PTC is how the system is designed, there is just too much beancounter driven "satisfice" with how things are implemented, which work great if they are fully working, but have little redundancy (it could be PTC has all kinds of fail safes, I don't know much about it, but I remain suspicious, safety to companies generally is seen as 'a cost', not a benefit, that hasn't changed much over time).
As far as technology being the cause of accidents, while I am not sold on self driving vehicles, especially in crowded areas (primarily to be honest because being a driver for now close to 40 years, I have seen how bad some people drive), no system is foolproof, and human beings are a lot more fallible than automated systems in a lot of cases. For example, road deaths number around 40,000 a year, and from experience with a rescue squad a lot of those are caused by human failings, DUI (even today), distracted driving (texting, fiddling with something, eating, you name it), or judgement calls (idiots in SUVs doing 70mph in snow and ice, because "I have AWD").A self driving car won't have most of these problems, they don't drink, they don't drop a cigarette or a hot coffee in their lap, they don't text, they don't stare at their navigation screen, or whatever.
They also don't get tired, driver fatigue, both in trucks and cars, is a major source of accidents. An automated system also will have much quicker 'reflexes' and can "see' much better than a human being can. If you look at what causes fatalities on the road, if you look at what causes accidents, there is no doubt that an automated system will remove a lot of them....so the real question is if an automated control system can handle not only road conditions, but also the reactions quite frankly of human drivers that don't always drive well, do unexpected things, and how well can the control system handle that, or other weirdness? What kind of certification testing is required on these control systems, and do you trust the corporate beancounters to follow them or game them (Volkswagon, anyone, with emissions certification/control? Would you trust a next generation Roger Smith to care about safety more or getting profits any way they can). My problem isn't driverless cars, as much as I love to drive, it is that I don't trust either the industry or the government to require the kind of things that make it safe. That said, though,I suspect once they are perfected (and they aren't yet IMO), while I am sure there will be accidents and yes, deaths, and some of them may be faulty technology or defects or breaks (then again, in a car today, brakes fail, steering joints break, transmissions fail, all of which can lead to accidents and deaths, especially if a car is poorly maintained), eliminating drunk driving, distracted driving, gabbing on the phone, eating, etc, will cause a huge decrease in accidents while the amount caused by tech failure will be a tiny fraction of that.
I don't think driverless vehicles are going to be common for a while and there is going to be resistance, both from people and for political reasons (with driverless trucks especially, with the loss of significant jobs, is going to cause pushback even assume they work correctly), but it eventually is likely to happen. Not sure how that will play out, if they will eventually ban manual driving, not sure. The real issue is going to be where jobs are involved, whether it be taxi or limo services, or trucking (or trains), rather than safety I suspect, technology has always supplanted jobs, the real question that no one has answered (and I am not going to,because a lot of it is political) is that with new technology it wiped out some jobs, craft workers replaced by factory hands as one example, it also tended to create a whole new score of others (for example, with mass production, producing parts on a mass scale at other factories, which involved labor), the real question is that with technology today, the new jobs created around the technology are being done by technology, too (put it this way, an automated factory requires a fraction of the labor of a traditional factory, and even if you factor in the people employed in the new factories, the people who create the technology (computer programming/support, network engineering, robotic tech), it is not likely to replace the jobs lost, pure and simple, there will be new classes of jobs created, as there was in the past, it is just that the numbers are not likely to add up and figuring out how to increase efficiency/safety/etc via automation but maintain a flowing economy, I leave that up to the experts.
Getting back to the original thread, train crews have been diminishing since there were trains. The air brake got rid of a lot of the brakemen jobs, with the advent of a diesel engine you no longer needed a traditional fireman or fully crewed helper engines bc of M/U, modern sensors and video has further cut down or eliminated the roles of brakemen and/or conductors, computer controlled and routed trains has eliminated a lot of the jobs associated with the actual shipping and making up/breaking down trains, containerized loads has changed things, too. I don't know if I like the idea of totally automated trains on any large scale, if they are controlled remotely, they can be subject to hacking and terrorism, and nothing automated is going to be 100% totally reliable or accident free, so having a person there makes sense. I will add might be a bit boring for the engineer if they are just the backup per se, the way I think that driverless cars might be.
