When the metal colleague gives a shove

Autonomous mobility as a further example of automation

Alongside robotics, autonomous mobility has also picked up pace again. Thanks to advances in areas such as sensor technology and, above all, artificial intelligence, the automation of mobility – which had previously stalled somewhat – has also been able to make progress. Vehicles with a high degree of automation are already in use in selected application areas, albeit mostly within clearly defined operational zones.

Strictly regulated routes on company premises can already be navigated today by autonomous shuttles without any human drivers at all. Here, too, logistics centres are leading the way. As they are under intense pressure to improve efficiency, they are managing with fewer and fewer human workers and are therefore an ideal testing ground for autonomous mobility.

In public spaces, however, the focus remains for the time being on intelligent assistance systems that relieve the burden on people but do not relieve them of responsibility. For here, too, the question naturally arises: who is liable for any accidents resulting in damage to property or personal injury?

The changing landscape of liability

Until now, liability issues have been regulated with relative clarity: if an employee operates a machine incorrectly or a driver causes an accident, responsibility can usually be clearly attributed to a single person.

With autonomous systems, however, the question of liability becomes significantly more complex. They operate on the basis of information from various sensor systems, which is processed by software. This creates a chain of action involving numerous parties: hardware manufacturers, software developers, operators, system integrators and data providers.

If an incident resulting in damage occurs, several questions immediately arise:

Was the sensor technology faulty? Did a software error influence the system’s decision? Was a faulty update installed? Or was the cause, after all, improper use?

Responsibility can thus shift from an individual user to complex technological systems involving various parties. For insurers, this ultimately means that traditional liability models are increasingly reaching their limits in this context and must be supplemented with new models.

The paradox here is that the drivers of technological innovation cite the reduction in accidents as a key advantage of autonomous systems. After all, machines do not make mistakes because they are overtired, distracted or lacking in concentration. Yet, as with almost every technological innovation, there is a catch: whilst human-caused accidents are usually isolated incidents, technical faults in autonomous systems can lead to cascading effects. A faulty software update or even a cyberattack on networked systems can affect numerous autonomous units simultaneously.

This creates a new liability landscape for insurers, as risks can also correlate with one another. Claims no longer occur in isolation, but can affect many systems. Furthermore, a faulty algorithm, for example, can also affect other system components and lead to further errors there. Just as with a complex mathematical calculation where an error occurs in the very first sub-task, all subsequent tasks will then also be incorrect.

Conclusion: How insurers can respond

The proliferation of autonomous systems is undoubtedly another major technological upheaval, much like industrialisation many years ago and, more recently, digitalisation. The insurance industry has always kept pace with these upheavals and adapted accordingly. This will certainly be the case again now.

One important approach lies in the development of new cover concepts specifically designed for autonomous robotic systems and vehicles. This is no longer just about traditional liability issues, but also involves the integration of product and cyber risks.

However, the innovation of autonomous systems also opens up new options and opportunities for insurers in risk assessment. Robots and vehicles generate a wealth of telemetry data for their autonomous functions, continuously providing information on their condition, usage and surroundings. This data can also be used to assess risks more accurately, analyse and investigate claims, and derive preventive measures.

Much like the digital immune system in IT [LINK], the insurance industry is also evolving here from a mere claims settler to an active risk partner. It is no longer simply a matter of insurance at the end of the value chain. Instead, collaboration with all stakeholders throughout the entire value chain is becoming increasingly important.

Text: Falk Hedemann