Tjaša Zapušek talked about differences between three different ‘decision-making processes’ that occur as a consequence of three different types of systems, namely, three different types of algorithms. Through this systematization she examined legal concerns regarding foreseeability and certainty of the system’s actions and applicability of existing legal rules. The increasing complexity of algorithms that are governing autonomous systems will result in systems with sophisticated self-learning capabilities, and more importantly, in the ability of creating unforeseeable and autonomous decisions. From a legal perspective, the test of foreseeability can be described with a sentence whether a person can see/recognize a systematic relationship between the type of accident that the plaintiff suffered and the defendant’s wrongdoing. Considering the aspect of foreseeability, a defendant may escape liability if scientists cannot predict a systematic relationship between wrongdoing and harm.

In the past couple of years, the model complexity has grown substantially. Some of these algorithms are often considered as ‘black boxes’, identifying that there is still lots of obscurity surrounding their operation. Even though the latter often rewards us with models of higher accuracy and productivity, it usually comes at the expense of human interpretability.

The main goal of this event was to prevent that this issue leads to either under or overregulation and to bring together experts with a background in law, technology, economics, sociology, and ethics to address the issue.