ABSTRACT
This concluding chapter synthesises insights from all contributions in Safety by Design, highlighting how automation, digitalisation, artificial intelligence (AI), and remote operations are reshaping safety-critical systems. While these technologies promise efficiency and sustainability, they also introduce new vulnerabilities when organisational readiness, human-centred design, and regulatory oversight fail to keep pace. Drawing on Rasmussen’s risk management framework and Human Factors research, the chapter argues that safety emerges from the interaction between technical, organisational, and human components, not from frontline behaviour alone. Across industries – maritime, aviation, nuclear, oil and gas, and transport – accident investigations demonstrate that treating ‘human error’ as a root cause obscures underlying deficiencies in design, organisational learning, and Management of Change. Psychological safety and a robust reporting culture are also essential for learning, enabling organisations to detect weak signals and manage emerging risks. Poor learning abilities negatively impact the implementation and exploration of automation and AI. Meaningful Human Control (MHC) becomes a central principle: humans must remain able to monitor, understand, and influence automated behaviour, particularly when AI or autonomy behaves unexpectedly. A Safety-by-Design approach must integrate three lifecycle pillars: 1) robust front-end engineering grounded in Human Factors, standards, and the EU AI Act; 2) operational practices that respect human limitations and build situational awareness; and 3) systematic organisational learning that aligns work-as-imagined with work-as-done. The CRIOP methodology is noted as a practical tool that unifies these dimensions by supporting design verification and operational validation and aiding accident learning. The chapter concludes that safe and sustainable automation requires continual investment in human-centred engineering, organisational readiness, and systematic learning to ensure that future advanced systems remain effective, trustworthy, and under meaningful human control.
