I’m straying to understand ways that safety functions executed using AI/ML technology could be validated. I’m looking for ways to check for deterministic behaviour. There is a project being worked on in IEC/TC 65/SC65A and ISO/IEC JTC 1/SC 42, but it’s a long way away from publication. The draft is pretty vague on many points. So, anyone here have any thoughts on this topic?

#AI #FunctionalSafety #SafetyFunctionValidation #Machinery