ABSTRACT
We have discussed in this book chapter how to give safety for artificial intelligence (AI) systems. If artificial intelligence is employed in safety-related applications, ‘we will take some AI model … for machine learning’. It shows statistical models. ‘In AI, use assessment normal procedure … on safety assessment model’. In this book chapter, we describe and give possible approaches for safety assessment of AI systems. Several questions remain open and are solved as separate applications. A SIL system determines E/E/PE system as normal. ‘This hazard is substantiated and risk analysed based’. It is very compulsory if the requirement of the system is not a SIL system. But in this case, AI can easily be used for this type of situation ‘if it has no occurred … a risk analysis’. ‘The AI system to integrate safety level needs … in AI system’. ‘We propose a challenging research decisive … to the AI systems’.
