Page 171 - Demo

time. Other issues, like the high cost of advanced AI models, resistance to organisational change, ethical considerations, and the need for robust data governance, add further complexity.Ultimately, education and training are becoming central to addressing many of these challenges. A more multidisciplinary approach is essential if the industry wants to remain competitive in this new environment.At the same time, more targeted policy support, continued investment in infrastructure, and greater standardisation of digital solutions will be essential to enable broader and more effective adoption of digital and AI-driven tools. Standardisation is particularly critical, yet it is not discussed enough and is often overlooked. Without standardisation, even the most advanced solutions may ultimately fall short of delivering real value.To what extent do data analytics and predictive models contribute to the reduction of fuel consumption and CO2 emissions on a dayto-day basis?Data analytics and predictive models are already making a meaningful difference in reducing daily fuel consumption and CO2 emissions. What makes them so valuable is their ability to support more accurate decision-making, enable real-time monitoring, and improve fuel management in ways that were previously impossible. I find hybrid models particularly interesting; by combining physics-based approaches with machine learning, they appear to strike a strong balance between accuracy and interpretability. Even with relatively limited datasets, these models can deliver highly reliable predictions, which is quite impressive. At the same time, they offer operational flexibility that enables continuous, real-time adjustments.However, it is important not to overlook what these systems depend on. Their effectiveness is closely tied to the quality and integration of data coming from multiple sources. Without well-structured and reliable datasets, even the most advanced models will struggle to deliver consistent results. In my view, their full potential remains somewhat constrained by ongoing challenges in organisational alignment, technical implementation, and data governance.Recent developments seem to reinforce the importance of hybrid approaches. Integrating domain-specific knowledge, such as hydrodynamics or engine-performance models, with machine learning is, I believe, one of the most promising directions forward. These models not only achieve very high levels of accuracy but also remain understandable, which is critical for real-world applications and trust.I see data fusion, AI, and real-time monitoring systems as key enablers of both operational efficiency and emission reduction. However, significant barriers remain to be addressed. Organisational integration is often more complex than anticipated, and issues around data governance and cybersecurity continue to raise legitimate concerns.May 2026 167