Lex Fridman – Playing Guitar in a Self-Driving Car

An evening pause: The guitar playing is great, but if this is supposed to be a demonstration of the abilities of self-driving cars, to me it is a utter failure. The drive was on a test track, with no other cars. The car itself was probably never going faster than 25 miles per hour.

In fact, if anything this proves the impracticality of self-driving cars. Such technology might work in a completely controlled environment, but as soon as you add any random human element, it can’t work. Thus our options: we continue to drive ourselves, or we give up our freedom to drive so that all vehicles can be autonomous.

But as I say, the guitar playing is great.

Hat tip Wayne Devette.

Software issue caused self-driving car accident that killed pedestrian

Sources in the investigation of an accident where a self-driving car hit and killed a pedestrian in Arizona in March now say that the car’s programming was at fault.

According to two anonymous sources who talked to Efrati, Uber’s sensors did, in fact, detect Herzberg as she crossed the street with her bicycle. Unfortunately, the software classified her as a “false positive” and decided it didn’t need to stop for her.

Distinguishing between real objects and illusory ones is one of the most basic challenges of developing self-driving car software. Software needs to detect objects like cars, pedestrians, and large rocks in its path and stop or swerve to avoid them. However, there may be other objects—like a plastic bag in the road or a trash can on the sidewalk—that a car can safely ignore. Sensor anomalies may also cause software to detect apparent objects where no objects actually exist.

Software designers face a basic tradeoff here. If the software is programmed to be too cautious, the ride will be slow and jerky, as the car constantly slows down for objects that pose no threat to the car or aren’t there at all. Tuning the software in the opposite direction will produce a smooth ride most of the time—but at the risk that the software will occasionally ignore a real object. According to Efrati, that’s what happened in Tempe in March—and unfortunately the “real object” was a human being.

I honestly do not understand the need for self-driving cars. In the end, I simply cannot see the software ever being capable of handling all the variables created by the presence of the unpredictable life that will surround it. And should it get that good, I wonder if we will then regret it.

German government sets ethical rules for self-driving cars

What could possibly go wrong? The German government’s Federal Ministry of Transport and Digital Infrastructure has established twenty ethical rules for the design and software of self-driving cars.

The German Federal Ministry of Transport and Digital Infrastructure has recently defined 20 ethical principles for self-driving cars, but they’re based in the assumption that human morality can’t be modeled. They also make some bold assertions on how cars should act, arguing a child running onto the road would be less “qualified” to be saved than an adult standing on the footpath watching, because the child created the risk. Although logical, that isn’t necessarily how a human would respond to the same situation.

So, what’s the right approach? The University of Osnabrück study doesn’t offer a definitive answer, but the researchers point out that the “sheer expected number of incidents where moral judgment comes into play creates a necessity for ethical decision-making systems in self-driving cars.” And it’s not just cars we need to think about. AI systems and robots will likely be given more and more responsibilities in other potential life-and-death environments, such as hospitals, so it seems like a good idea to give them a moral and ethical framework to work with.

It appears these geniuses came up with these rules based on a “virtual study.”

In virtual reality, study participants were asked to drive a car through suburban streets on a foggy night. On their virtual journeys they were presented with the choice of slamming into inanimate objects, animals or humans in an inevitable accident. The subsequent decisions were modeled and turned into a set of rules, creating a “value-of-life” model for every human, animal and inanimate object likely to be involved in an accident.

“Now that we know how to implement human ethical decisions into machines we, as a society, are still left with a double dilemma,” says Professor Peter König, an author on the paper. “Firstly, we have to decide whether moral values should be included in guidelines for machine behaviour and secondly, if they are, should machines act just like humans?”

I know that my readers will immediately reference Isaac Asimov’s three laws of robotics, but that really doesn’t work. Asimov’s laws were incorporated into a science fiction “positronic brain” that was supposedly built almost organically, so complex in formation no one really understood it. Once the laws were incorporated into each brain they could not be tampered with without destroying the brain itself. Our coming robots will have no such protection.