THE NEXT LEVEL OF MOBILITY

Dr. Christoph Endres: Stefan, I’m delighted that you’re here today. Your company Veecle is considered to be particularly progressive. Could you briefly explain what exactly you do?

Dr. Stefan Nürnberger: We offer a software development kit that enables vehicle manufacturers – whether buses, scooters, cars, trucks or agricultural machinery – to program the vehicle functions themselves, which are usually defined via control units from suppliers. This decouples the logic and software from the actual hardware. Effectively, we enable the development of software-defined vehicles in which all functions can be programmed in advance without having to commit to specific hardware at an early stage. The final hardware can be selected on short notice, right before the start of production.

Dr. Christoph Endres: In the last year and a half, the topic of generative AI has become extremely popular. How do you think it will change the automotive market or the way we handle our vehicles?

Dr. Stefan Nürnberger: Generative AI, especially large language models, significantly improve the understanding of human instructions. For example, the car could respond to the statement “I’m cold” by recognizing who said it – whether driver or passenger – and adjusting the air conditioning or closing a car window accordingly. This represents a significant advancement over previous speech-to-text applications and helps to reduce the cognitive load for users.

Dr. Christoph Endres: Will interaction with the car be limited to us giving it instructions, or will it also be able to make small talk?

Dr. Stefan Nürnberger: In the future, it will be possible for the car to engage in small talk. Some manufacturers are already experimenting with explainable AI, for example to make the vehicle’s decisions transparent to the occupants – such as why it brakes and changes lanes. This helps occupants to understand the car’s actions and increases trust in the technology.

Dr. Christoph Endres: How do you rate car manufacturers’ awareness of the safety risks of voice models?

Dr. Stefan Nürnberger: It varies from group to group, but in general the automotive industry tends to be generally risk-averse and hesitant when it comes to introducing new technologies. They wait and see whether and which safety risks exist, especially as functional safety is also involved. Connectivity comes with new security vulnerabilities, and if it’s not properly implemented, malware and other threats can penetrate through them.

Dr. Christoph Endres: Do you think that blackmail attempts could be made while driving, such as forcing immediate payments? For example, with risky driving maneuvers, refusal to continue driving, etc.?

Dr. Stefan Nürnberger: I don’t know of any real cases, but that would theoretically be an effective scenario. However, it is unlikely that a Bitcoin transfer could be carried out quickly enough under such conditions.

Dr. Christoph Endres: What kind of damage could theoretically be caused by Large Language Models (LLMs), especially if they don’t rely solely on voice input?

Dr. Stefan Nürnberger: The damage could be manifold, including reprogramming of the navigation system or unintentional control of other vehicle functions through hidden commands. This could happen if the vehicle has access to the Internet and uses external sources. There is a clear need for further security measures in order to minimize such risks.

The conversation with Dr. Stefan Nürnberger offers a fascinating insight into the future of automotive technology and the role that Generative AI could play in it. Thank you for the interview.