Wolfgang Lehmacher writes for Splash today, making the case for protecting new recruits to the shipping industry. Generative AI can be the productivity booster of the century for shipping and supply chains. However, it can also shrink the talent pool. Economist Enrique Ide’s paper ‘Automation, AI, and the Intergenerational Transmission of Knowledge’ argues that by aggressively automating early career work, we risk trading short-term efficiency for long-term decay in human capability.​ Ide’s model centres on tacit knowledge, the hard-to-codify skills people gain through real-world challenges early in their careers. AI reduces the spaces where juniors traditionally learn, creating what he calls “socially excessive automation” of entry-level tasks. His estimates suggest that aggressive entry-level automation could shave 0.05 to 0.35 percentage points off long-run annual per-capita growth; seemingly small, but cumulatively large when compounded over decades.​ In supply chain and logistics, tacit knowledge can be the difference between a nearmiss and a week-long shutdown. Dispatchers sense that a port is about to clog. Planners know which supplier buckles under pressure, and warehouse supervisors spot safety risks based on experience. These capabilities are forged in long days and endless rounds of trial and error.​ Today, machine learning engines generate demand plans and replenishment proposals that junior analysts once built line by line. In logistics, optimisation tools route trucks and containers with minimal human intervention, reducing the need for coordinators to wrestle with cut-off times and last-minute changes.​Generation AI may know dashboards better than operations. In many automation roadmaps, the first targets are precisely the repetitive, data-heavy tasks that historically gave juniors their early exposures and repetitions.​ Supply chains are uniquely exposed because they are continuously stress tested by pandemics, canal blockages, labour disputes, wars, port congestion, or natural disasters. When digital tools fail because data is missing or the world no longer resembles the past, the system falls back on human judgment. If that judgment has never been built, resilience fractures. Today, planners start losing the ability to think holistically, young managers struggle with numbers, and AI programmes target the very tasks where those muscles once developed.​ Ide proposes policies that subsidise apprenticeships, tax entry-level automation, and promote AI that complements rather than replaces junior work. This aligns with a broader research consensus: the highest returns from AI come when organisations design for humanAI teaming, not full automation, especially in complex, uncertain environments like supply chains. Systems that combine humans and AI outperform automation for judgment-intensive decisions, but only when organisations engineer those interactions.​ The call to action is clear. First, treat early career learning capacity as a strategic asset, not a cost. Preserve and redesign rotational programmes, control tower war rooms, and on-site assignments where graduates sit beside seasoned operators and make real decisions on live flows. Second, make apprenticeship by design a non-negotiable in AI deployments. When AI automates a task, organisations should ensure juniors can see, interrogate, and override the system, especially in exceptions management, supplier risk assessment, and scenario planning. Value in supply chains