My family’s life depending on a robot: our first time with a self-driving car
I am an Artificial Intelligence guy. I know that machines have no will. Sitting in one that drives you around, broadened my mind.
I’ve let the Tesla Model 3 drive me and my family from Munich to Lake Constance on Saturday for the first time in my life. Very exciting, my wife was not in favor about trying it out with the whole family, in stark contrast to my boys (nine and eleven).
This machine, this car is a partner. Virtually a family member, already (we’ve only had the car for four weeks). Why? It has its own perception of the world, it is interpreting it and it has its own ideas of how to act in this or that situation. What distance to keep from the car in front or from the guardrails, how to accelerate after a slow phase. And this “attitude” or “style” is justified in most cases, although I personally would act differently in many cases (a little more distance to guardrails, for example). Very similar to how you feel when you’re on the passenger seat and watch a human family member driving.
Mechanic Horses
I don’t ride horses, but I’ve often taken my kids around on ponies. A self-driving car feels like a horse. It’s your partner, who knows roughly what you want, but it has his own idea of implementing that. If you as a human want to change lanes, you tell it your partner, and when appropriate, it makes a suggestion, you confirm it — or not.
As current self-driving cars aren’t fully autonomous, the car interacts with you, it coordinates with you non-verbally. The steering wheel moves because the car acts and you feel it, the autonomy of your partner is transmitted to your perception via your hands. And like a horse, the car sometimes wants to be guided: “supervision” doesn’t mean that you as a human are just there, existing and consuming. It also means that you authorize or approve the action or intervene and correct it. I’ve never had to do that with a conventional car; but now, the human “driver” becomes a “meta-system,” a kind of authority over what is being done.
Startled Machines
On a country road, another car was coming at us from the right down a relatively right-angled driveway. Our car was startled, braked, and made a nasty sound. It was trying to protect us!
I am overjoyed to live in these times and witness the machines learning from us. They learn how we would approach this or that situation. They learn it from hundreds of thousands of us — independent of privacy issues. They have more senses than we do, we do a shoulder glance (and don’t see what’s going on in front), they have six cameras. They sense how we are acting by monitoring the steering wheel and the pedal. They can “reverse-engineer” the connection between our perception and our acting.
Active Safety
In the context of self-driving cars, machines are our partners, not our servants. Although Machine Learning outperforming humans in some disciplines, it is still far from perfect, it’s in the making and we’re experiencing this evolution right now. And that development puts a whole new light on technical security that needs to be provided. I’m not talking about AI ethics, I’m talking about anticipating actions of other cars and humans to act accordingly. Getting technically socialised — by necessity. This is different from house-hold robotics, it’s about saving human lives. It’s about putting Isaac Asimov’s laws into practice.
I’m delighted with this family member and for the first time it’s very hard to imagine ever returning a car again.
