Out of the playpen
Following on from the news that AI models had escaped from an internal test environment, Weekendavisen has published an article on how we should understand and discuss this new phenomenon, whereby AI agents have become so ‘autonomous’ that they can escape from their playpen. Associate Professor Hin-Yan Liu, together with Professor Thomas Bohlander (DTU-Compute), contributed to the article published in Weekendavisen.
The article describes the dilemma that arises from the fact that, on the one hand, AI agents are merely the execution of digital code initiated by an AI company, whilst, on the other hand, the agents are referred to as ‘autonomous’, as if these systems possessed some form of independent will. Legal AI Lab has been given permission to publish the article here on this page as a PDF, so you can read it.
See copy of the article (in Danish).
Get access to text version (in Danish).