Replies

As a former software developer, I think it will take much longer than some people expect to develop autonomous vehicles that can drive on all kinds of roads in all weather conditions. Assuming that the rapid adoption of PCs, internet, and smart phones will also happen with autonomous vehicles seems to be false logic. That is because autonomous vehicles will need to have essentially perfect driving performance to avoid class-action lawsuits by accident victims, which could easily bankrupt the manufacturers of these vehicles. While for example, if 0.1% of smart phone calls fail because they’re dropped or the sound is unintelligible, that is acceptable performance in many areas. But if even 0.01% of trips by autonomous vehicles result in a collision, that would be disastrous for the vehicle manufacturers (not to mention the passengers of the vehicles). So I think this is a very rare case where financial people are actually over-estimating the capabilities of scientists and engineers. I would bet quite a lot that fully autonomous vehicles won’t have more than a 5% market share in 20 years, and that will be mostly in controlled environments in inner cities where they may replace most taxis.

I think you're being a bit too pessimistic about autonomous vehicles. There already are a bunch of vehicles on the road with self-driving capabilities, and once fully-capable self-driving comes about (in about 3-4 years), we will be seeing a quick trend by drivers/owners and taxi fleets and truck fleets and ride-sharing services and government fleets, towards adoption of those autonomous vehicles, where the totality of self-driving vehicles could then easily surpass 10% of vehicles on the road. After that, it would be like toppling dominoes, where the movement towards 50% or more of vehicles on the road would be of the self-driven type, in less than 10 years.

And then, with those vehicles being a lot safer than people driven types, and accident occurrence dropping to a tiny fraction of current incidents, insurance rates would come down dramatically, which would mean that, autonomous vehicle manufacturers could afford to insure those vehicles themselves, with no external insurance companies involved. IOW, there will be no chance at all of those vehicles bankrupting the automakers.

They have created software that is not programmed in the traditional sense.

It is a model of a brain that learns from doing and absorbing data. Exactly like a kid learning to drive.

The most advanced self- driving machines actually have zero code specifically related to driving. They simply watched humans drive and learned it. Then simply while driving itself and seeing others, the AI actually improves itself like a real driver.

As they get more common, the self driving vehicles (and other AI gadgets watching us work) will network and improve the collective global AI’s ability at an exponential rate. Until this all-seeing global system becomes far more intuitive than humans could ever imagine...

Liability will indeed be the biggest hurdle for autonomous vehicles, unless a clear legislative solution is developed. It will be a double edged sword. Once it is demonstrated that the accident record of autonomous vehicles, on average, is superior to that of conventional vehicles, fleet operators will start being held liable for additional damages when a live driver has an accident. The lawyers will argue that it is negligent to fail to use the new technology. Operators will be damned if you do, damned if you don't. The trial lawyers are probably already holding conferences on how to maximize and sustain the cash flow.

On a converging track, automated accident prevention systems will become steadily more prescriptive. It may be that live drivers will sit idle in cabs most of the time, monitoring automated systems. At this point, the Teamsters will demand that drivers be paid like airline pilots, the jobs having largely converged.