Prof. dr. Edward A. Lee, UC Berkeley (USA)
To celebrate the achievements of IPN in the past 25 years, IPN is organising a special, online series of colloquia in which world-renowned computer scientists give their view on the progress in, and future of the field of computer science. These colloquia will feature thought-provoking presentations that are of interest to a broad (academic) computer science audience. Although these colloquia are initially aimed at the Dutch computer science community, they are open to interested people around the world!
Abstract
Cyber-physical systems are systems that integrate software and networking with physical components that sense and actuate in the physical world. Traditionally, to make them trustworthy, engineers strive to make their behaviors predictable, repeatable, and provably safe. While security is essential to achieving these objectives, it is by no means sufficient. Many technical challenges arise with sporadic connectivity, real-time behaviors, and handling of inevitable faults. While the integration of machine-learning-based AI systems holds promise of being able to help deal with many of the security and other challenges, they simultaneously make behaviors less predictable and repeatable. In this talk, I postulate that the certainty that is traditionally achieved with rigorous engineering methods may be fundamentally incompatible with integrating intelligence into our systems. The question then becomes, how do we get the advantages of intelligence with acceptable risk?
Date, Time, Location
The colloquium will take place on September 18, 2025, 16:00 – 17:00 (CEST).
The colloquium will be hosted as a Teams webinar. You can find the connection details on the IPN website: https://ict-research.nl/ipn-colloquia/ or join the colloquium directly here.
We are looking forward to seeing you on September 18th!
Afterwards you can find the recording here.