Existential Intelligence: dissecting AI danger and our (in)abilities to control it - Roel Dobbe (ENG)
LOCATION: GRAVENSTEEN
LABEL: TALK
TIME: 19:30
Discover how society and our elites systematically underestimate the effects of artificial intelligence and our control over it, and what we can do about it.
Recently, the question has arisen as to whether artificial intelligence (AI) could lead to existential scenarios where humanity is wiped out. The rise and enchantment of ChatGPT and similar AI features has not only given doomsayers the ammunition to see their theories of superintelligence confirmed. Tech industry and political leaders have also eagerly responded to these wonderful doomsday scenarios known from Hollywood films and time-honored science fiction novels and legends. But are those concerns justified? Does this frame perhaps miss the existential scenarios that are already unfolding? And what can we learn from history to manage advanced automation?
Roel Dobbe is a researcher at TU Delft, at the Faculty of Technology, Policy and Management. He leads an interdisciplinary research group that investigates the risks of complex algorithmic systems in critical infrastructure, including the energy system, social security and healthcare. They try to explain across domains and shape what is needed to ensure safety and justice, both in the technological development of AI systems and in the organization and management of processes that rely on this rapidly developing technology. It sounds complex, but the crux is often very simple!