Artificial intelligence is increasingly capable of automating professional tasks traditionally performed by entry-level staff. While such automation may drive short-term efficiency and profitability, it risks disrupting the experience-based learning pipeline essential for developing a firm's long-term innovation capacity. This paper introduces the Knowledge Trap: a structural condition where immediate gains from AI adoption lead to the gradual degradation of the tacit knowledge and adaptive search capabilities required for future performance. Using an NK agent-based simulation model, the study examines how different sectors and governance horizons influence this risk. The model distinguishes between three adoption strategies - substitutive, hybrid, and complementary - to track their long-term effects on a firm's experience pool and innovation fitness. The findings demonstrate that the Knowledge Trap is most severe under substitutive adoption, particularly in sectors with high task interdependency. Conversely, adopting AI in a complementary mode can substantially mitigate these capability losses. This research contributes a forecasting framework to help organisations and policymakers identify structural risks and develop AI implementation strategies that protect the human talent pipeline.
A directed phylogenetic network is tree-child if every non-leaf vertex has a child that is not a reticulation. As a class of directed phylogenetic networks, tree-child networks are very useful from a computational perspective. For example, several computationally difficult problems in phylogenetics become tractable when restricted to tree-child networks. At the same time, the class itself is rich enough to contain quite complex networks. Furthermore, checking whether a directed network is tree-child can be done in polynomial time. In this paper, we seek a class of undirected phylogenetic networks that is rich and computationally useful in a similar way to the class of directed tree-child networks. A natural class to consider for this role is the class of tree-child-orientable networks which contains all those undirected phylogenetic networks whose edges can be oriented to create a tree-child network. However, we show here that recognizing such networks is NP-hard, even for binary networks, and as such this class is inappropriate for this role. Towards finding a class of undirected networks that fills a similar role to directed tree-child networks, we propose new classes called q-cuttable networks, for any integer q >= 1. We show that these classes have many desirable properties, similar to tree-child networks in the rooted case, including being recognizable in polynomial time, for all q >= 1. Towards showing the computational usefulness of the class, we show that the NP-hard problem TREE CONTAINMENT is polynomial-time solvable when restricted to q-cuttable networks with q >= 3. (http://creativecommons.org/licenses/by/4.0/).
Model-based engineering (MBE) is a powerful paradigm that leverages models as essential pillars of the development process, enabling teams to clarify requirements, streamline design, specify behavior, and perform rigorous verification and validation tasks across the entire system life cycle. Digital twins (DTs) represent revolutionary software systems that mirror cyber-physical, socio-economic, or biological entities, systems, or processes. Built from robust models and data, DTs are deployed for high-impact applications such as planning, monitoring, control, and optimization of the twinned entity. The model-centric nature of DTs has naturally ignited recent exploration into harnessing MBE for the engineering and operation of DTs. However, this organic evolution has created a fragmented landscape of (partial) solutions. To confront this challenge, this article presents a rigorous and systematic literature survey on the field of model-based DT engineering (MBDTE), accompanied by a novel taxonomy for categorizing MBDTE approaches. We also introduce crisp definitions of both the field of MBDTE and the models themselves. We conclude by highlighting research gaps and outlining avenues for further exploration.
Whilst performance analysis is a key component of sport science and coaching practice, the novelty of this research is its focus on the often-overlooked skills and competencies needed to operate effectively within applied environments. The study explored how performance analysts collaborate with coaches and athletes in high performance sport, emphasising the relational, practical and technical competencies that underpin successful practice. Thirty-eight (38) performance analysts (experience 7.2 +/- 3.8 years) in the elite UK performance analysis system completed semi-structured interviews (24 questions, lasting 38 +/- 12 min), which were analysed using Braun and Clarke (2006) six-step approach. Findings highlighted that effective performance analysis relies on technical expertise, practical skills, and relational competencies. Trust, openness, and credibility were central to collaboration, while technical skills such as data collection, analysis, and tailored feedback ensured outputs were accurate and actionable. Analysts navigated organisational, technological, and resource constraints, with soft skills including adaptability, resilience, and communication critical to overcoming barriers and achieving applied impact. A tree analogy illustrates the main thematic areas ([1] data and video, [2] analysis, [3] feedback, [4] trust and openness, [5] knowledge, [6] culture, and [7] challenges), their sub-components, and the interwoven soft/hard skills underpinning effective daily practice. These findings have clear implications for educational providers and practitioners across all career stages, offering practical guidance to enhance applied practice.
