2nd International Workshop of Artificial Intelligence Safety Engineering (WAISE) was held at September inTurku. The goal of this workshop is to bring new ideas to the domain of AI safety. The presented works range from philosophical answers to practical solutions. FIMA was represented by doctroral student Saeed Bakhshi Germi . He made a summary about selected papers, and got contacts with the globally leading research in this domain.
An interesting note is a preliminary draft of UL 4600: Standard for Safety for the Evaluation of Autonomous Products. It is evident that new technology and requirements are not entirely covered in existing standards. The new standard is planned to be an additional standard on top of ISO 26262 and ISO/PAS 21448 to solve the issue.
Another interesting topic was how to improve ML safety with Partial Specifications. The presentation reviews over types of partial specifications that are applicable for ML.
Saeed´s report and links to the most interesting papers are in the login pages.