When programmed correctly, computers almost never fail at their assigned task. Humans, on the other hand, can and do make mistakes. I think if we reach a point where automated cars are programmed to react to all possible contingencies, they will almost certainly be safer on the road than humans, especially tired ones.
Autonomous vehicles have driven for millions of miles over the past 5 years with astonishingly safe results- not a single accident. It's not just that autonomous vehicles don't get tired. Unlike humans, robotrucks don't have blind spots and are not blinded by fog or rain. Autonomous vehicles are able to calculate brake-to-weight ratios with precision and closely monitor the temperature of brake pads, eliminating the danger of runaway trucks. A robotruck will never try to drive under a bridge with insufficient clearance or into a flooded roadway where the depth cannot be gauged. There's no such thing as a flawless system and I'm sure that when a robotic vehicle fails it will seem more ominous than human failure. However, the safety record will make autonomous vehicles hard to refuse. Say, just by way of example, that Target uses only robotrucks, has one accident a year and no fatalities while Wal-Mart uses humans, has 50 accidents killing 5 people per year. How many years can Wal-Mart sustain human drivers before the lawsuits become unmanageable? Truckers killed 4,000 Americans in 2012. If technology can reduce that number to a handful or likely even zero, how long will it be before Americans demand that kind of safety? Many of the largest haulers of US goods are already making plans for conversion to autonomous vehicles and insurance companies expect to see a mass conversion by 2025. Google is predicting that autonomous vehicle programs in conjunction with drive sharing programs (like Uber) could reduce the cost of transportation by 90% by 2030. There is no doubt that autonomous trucks are exponentially safer than human drivers or that robots will replace drivers in the near future. The only questions are how quickly can vehicles, laws, insurance coverages be converted to the new paradigm and how resistant will America's car culture be to the change?
I believe this world will not learn how dangerous computers can be until a major catastrophic event occurs at fault of a computer. Computers always stand a chance of crashing or other failures, and I for one do not, and will not ever trust a computer to drive my family around. I don't even want to be sharing the road with a computer driven vehicle. I think computers operated trucks will make people less intelligent behind the wheel when they learn to rely on the computers for too much.
Although I think it is a great idea, I would not be comfortable just having a computer drive a truck. What would happen if the computer would just stop working and there would be no one there to stop it. I think that although in theory it is a great idea technology is not there yet for it to become a option
No, computers are not safer drivers on the roads than tired truck drivers, because there are too many ways that computers can go wrong. It only takes one hacker to mess with all the trucks and hurt and kill a lot of people at once. We are still better with using human intelligence on the roads.