People are left to themselves-driving cars in the future far-reaching decisions – about who dies in a collision. Simulates the in an Online Experiment, this brings us to moral limits.
The mit Media Lab
More articles by Christoph Stockburger
- Rights of use
A woman and two children sitting in a self-driving car. Suddenly the brakes fail. The car is racing directly at two men and a woman, walk straight across the road. What happens now: Either the three pedestrians from being run over or the car swerves against a concrete wall, and the three occupants of the car die.
Who should live, who should die – how would you decide?
The question comes from the Experiment of the morality Machine. You can go online through any number of times, for a total of 13 of such scenarios are available for each. Behind the project, the Media Lab of the Massachusetts Institute of Technology (mit), which focuses on the exploration of digital technologies.
The researcher Iyad Rahwan has designed the morale of the Machine. His personal decision on the above-described scenario with the car and the pedestrians, he does not want to reveal after a brief consideration, dear.
The mit Media Lab
Who’s going to survive the Crash, who is to die? The morality Machine provides the participants with difficult questions
SPIEGEL ONLINE: Mr Rahwan, the morality Machine seems playful but the scenarios are a disaster: You have to determine who is in a car accident dies and who survives.
Iyad Rahwan: Yes, one of these moral choices an unpleasant feeling. But we want to provoke an ethical debate about self-driving cars. Because it helps the subject to it emotionally.
To The Person
Iyad Rahwan is a Professor at the mit Media Lab. He deals primarily with computer technologies, as well as artificial intelligence, and examines their social aspects. Most recently, he has written together with Jean-François Bonnefon, and Azim Shariff, a study on “The social Dilemma of Autonomous vehicles”.
SPIEGEL ONLINE: In the scenarios are not only the age and gender of the people, but also their professional and social Background. It is robber is homeless, a Manager or a Bank. Why does it matter?
Rahwan: You should artificial intelligence do not underestimate, she is to pretty scary things. Algorithms can already detect very quickly and reliably faces. But of course, no manufacturer would build such a feature in his car, that would be wrong.
SPIEGEL ONLINE: Why, then, the classification of the Moral Machine?
Rahwan: Because self-driving cars any passers-by, at least theoretically, a certain social importance, in order of which decision to make. People of such Considerations in their Actions to influence them. The user of the morality Machine should be reminded.
SPIEGEL ONLINE: Can you give an example?
Rahwan: The children are worth protecting, because you in comparison to the elderly, have a longer life expectancy. The other holds the Old is more important, because you can give life experience to the Younger. Some equate the social significance of people and animals. You will need to decide on our project about death and life, one is confronted with these questions.
The mit Media Lab
In the case of the morality Machine, it is not only human but also animal life
MIRROR ONLINE: the morality Machine to learn more about yourself than about self-driving cars.
Rahwan: Exactly. At the same time, we want to make it clear that the use of artificial intelligence more and more compromises must be made: people need to ask themselves what their morality should ideas take over the car’s system and which are not.
Prototype of self-driving cars from Google
SPIEGEL ONLINE: Account of the auto industry, these ethical issues in the development of Autonomous vehicles?
Rahwan: The manufacturer in the discussion. You are engaged in self-driving cars so far, especially with technical challenges, legal framework and insurance questions.
SPIEGEL ONLINE: have to the question of how Autonomous vehicles react in certain situations, not in front of a possible road licensing be clarified?
Rahwan: The discussion should at least not be moved. At the present time is open, who is to decide how a self-driving car in a given Situation. If you leave this decision solely to the manufacturers, they set the priority likely to be on the safety of vehicle occupants – and that could be bad for the Rest of the population, for example for pedestrians. However, it is important that the interests of all road users be taken into account.
(Click here to read a detailed article on the attitude of the car industry to the ethical aspects of Autonomous driving)
Push messages personalized
You are a car enthusiast, love bikes, or excited to follow the mobility change? With the new sections added to the Alarm in the SPIEGEL ONLINE App you miss on these topics, not articles. Here, you will learn how to off and on.
SPIEGEL ONLINE: The people must have the Power of the machine. Do you have a comparable example?
Rahwan: you have to Take the so-called Filter Algorithms in social networks: it is the decision of an artificial intelligence, what are the content and advertising displayed to the users. It learns from the user behavior and optimizes accordingly. This is on the one hand, practically. On the other hand, the Hiding of dissenting opinions seems polarizing, there is the risk that you will notice the different arguments of a debate. That individual Algorithms have an impact on the whole of society, we will see in the future more often. Thus, a great responsibility goes hand in hand.
SPIEGEL ONLINE: What does this mean concretely, in the case of self-driving cars?
Rahwan: the manufacturer, the cars are not programmed to be allowed, that the life of the inmates is a priority. And, above all, that the authorities need to check the manufacturers ‘ Algorithms.
SPIEGEL ONLINE: How is that possible?
Rahwan: The programming codes for self-driving cars are so complex that independent auditors the commands hardly see through. However, one can observe the behaviour of these vehicles in certain situations and analyse them.
The car service Uber has been testing self-driving cars in Pittsburgh, cooperation partner of the manufacturer Volvo
SPIEGEL ONLINE: it sounds as if the cars a life of its own.
Rahwan: The Software of self-driving cars, though people are programmed to learn on their own. The programmer will make the car a target, for example, to drive comfortably or safely and an algorithm ensures that the car comes from experience and from their own data analysis, this goal is getting closer and closer. On the basis of the Codes, this artificial intelligence is to understand sometime. So, to avoid that there are unwanted effects, for example, that the car takes less consideration to pedestrians, needs to be analysed and its behaviour exactly. As in a human.
SPIEGEL ONLINE: You have the cars like people treat?
Rahwan: In a certain way. To check whether a person adheres to social rules, you can’t look into his head – you can see it only in its Action and, where appropriate, to misconduct. And this is, in my view, the most effective method in dealing with artificial intelligence.
SPIEGEL ONLINE: If cars are in essence as unpredictable as people are, the roads not just safer.
Rahwan: Oh, no, I am quite of the Belief that machines such as self-driving cars have the potential to have less errors than people. We should be aware of – and here we are again with the Moral Machine, – that even the best technology has to sometimes make compromises.
SPIEGEL ONLINE: look Forward to the day when the cars of race alone?
Rahwan: If there are less accidents there, should the people forever from the steering Wheel to be banned.