AI is the key

AI is crucial to autonomous driving, but there are still many questions to resolve.

Through the beyond Initiative, Audi is ensuring that AI is used for the benefit of society as a whole. Project Leader Barbara Wege, advance development engineer, Dr. Miklós Kiss and lawyer, Martin Siemann, discuss future challenges and benefits.

12 September, 2018


Artificial Intelligence (AI) is a key technology for automated driving, and will also bring substantial changes to the way we work. Three key people (pictured above left to right: Lawyer Martin Siemann, advance development engineer, Dr. Miklós Kiss and Beyond Project Leader Barbara Wege) spoke to Audi Magazine about just what it will take to make it happen and what it could mean to society as a whole.

Audi Magazine: Barbara, can you briefly outline the purpose of the beyond Initiative

Barbara Wege: The initiative assumes that artificial intelligence will have an increasingly profound influence on our mobility, on the way we work and on our lives as a whole. Our mission is to play our part in ensuring that AI is beneficial to our society. To this end, we have set up an interdisciplinary network within which experts from different Audi functions, including, for instance, data scientists, sociologists, psychologists, philosophers, economists and legal experts. 

Audi’s main interest, at least initially, is surely autonomous driving… 

Barbara Wege: Yes, and above all, the ethical, legal and social aspects. Beyond that, we’re looking at how AI will influence the way we work. 

Miklós, you’re not only a technical expert, but also a specialist in human perception and communication. What reservations do the public have in respect of AI in cars – and how can they be addressed? 

Miklós Kiss: Reservations always exist where people fear a loss of control. We are introducing automated driving step by step – partly in order to build up trust within society and politics. The new A8, which has the capability of driving autonomously on Level 3, and the Audi Aicon show car are part of that process. 

Martin Siemann: It’s crucial that we encourage social acceptance of the technology. And in respect of the legal regulations currently being established, it’s also important that society sees them as appropriate and embraces them accordingly.

Reservations always exist where people fear a loss of control

When automated driving it discussed in the public domain, it’s often within the context of the so-called dilemma situation, which is a scenario in which an automated car can no longer avoid an accident and can only choose which of two pedestrians to hit. How relevant is this problem to your work? 

Miklós Kiss: The dilemma situation is first and foremost a textbook example, and can’t be solved by our ethics and our constitution. In an unavoidable situation, when a human being or a machine has to decide between two options, one of those will prevail. Fortunately, while such situations are conceivable, they are extremely rare and don’t feature in accident statistics. As automakers, it can’t be up to us alone to decide the way autonomous cars should react to such a situation in future, which is why the discussion in the beyond Initiative is so important. As developers, what we also take from the dilemma debate are other related questions such as: Should a car avoid a collision if, for instance, it can’t see where that avoidance manoeuvre would take it? Should it drive into a garden fence when there might possibly be a person behind it? 

Martin Siemann: And should we program a vehicle to be able to break the law in an emergency situation? For instance, to cross a solid line during an avoidance manoeuvre? Should it be able to drive through a red light if an ambulance is approaching from behind? These are questions that must be addressed by automakers and legislators. 

In an unavoidable situation, when a human being or a machine has to decide between two options, one of those will prevail

As things stand, we don’t use self-learning systems in any situations that are a matter of safety

In future, will cars continue to learn as they drive and be able to make these and other decisions independently? 

Miklós Kiss: As things stand, we don’t use self-learning systems in any situations that are a matter of safety. In the medium term, we will combine artificial intelligence with conventional technologies in order to offer maximum safety and to reinforce customers’ trust in the technology. 

Martin Siemann: AI in vehicles capable of developing themselves will cast up a completely new set of questions, mainly with regards to manufacturer liability. Some of our colleagues are already considering whether such vehicles should be defined as electronic legal entities – although anything like that is definitely several years away.

Experts predict that autonomous driving can avoid 90 percent of all accidents. Will the four of us here live to see the day when present-day manual driving is prohibited? 

Miklós Kiss: In the far distant future, I can very well envisage manual driving being prohibited on freeways and in city centres. On out-of-town roads, which is the most complex environment, manual driving will still be necessary for a very long time to come. In the past, every form of mobility that ever existed has been preserved as a sport – running, horse riding, swimming. The same will apply to motorsport. 

Martin Siemann: The switch to autonomous driving won’t happen through prohibition alone, but also through incentives. These could be the added safety, lower insurance payments or fast lanes on freeways and motorways for self-driving cars.