Emerging targeted herbicide application technologies, such as the ONE SMART SPRAY™ system (Bosch BASF Smart Farming, Cologne, Germany), enable postemergence herbicide application only where weeds are detected. Field studies were conducted at four site-years in Seymour, IL, and Janesville, WI, during 2023 and 2024 to compare area treated and waterhemp control between broadcast and targeted herbicide applications in soybean. In Seymour – 2023, early postemergence application (V2 growth stage) resulted in less area treated (26%) than the standard time but required inclusion of a layered residual herbicide to reach 89 to 93% waterhemp control and biomass reduction ≥94% at 28 days after standard application time (DASAT). The standard time (V4 growth stage) in Seymour – 2023 provided 87 to 97% waterhemp control and ≥94% biomass reduction at 28 DASAT but resulted in more area treated 76% at postemergence application. In Seymour – 2024A and Seymour – 2024B, where waterhemp densities were low (≤5 plants m-2), treatments had >96% waterhemp control and biomass reduction 28 DASAT, and targeted applications resulted in 17 to 34% area treated, regardless of application time. In contrast, in Janesville – 2024 (high waterhemp density; > 30 plant m-2), waterhemp control and biomass reduction did not exceed 90% at 28 DASAT, where area treated was ≥97% at postemergence application, indicating that targeted spraying essentially functioned as a broadcast application and would have required additional postemergence application(s) for effective season-long control. Under high waterhemp densities, frequent nozzle activation reduced potential herbicide savings and diminished differences from broadcast applications, clearly indicating that targeted systems are best suited for fields where effective foundational management practices, including robust preemergence herbicide programs and other integrated weed management tactics, reduce weed density during postemergence application season.

Skill extraction is at the core of algorithmic hiring. It is based on identifying terms commonly found in both targets (i.e., resumes and job offers), aiming at identifying a “match” or correspondence between both. This article focuses on skill extraction from resumes, as opposed to job offers, and considers this task both from the human resource management (HRM) and artificial intelligence (AI) points of view. We discuss challenges identified by both fields and explain how collaboration is instrumental for a successful digital transformation of HRM. We argue that annotation efforts are an ideal example of where collaboration between both fields is needed and present an annotation effort on 46 resumes with 41 trained annotators, resulting in a total of 116 annotations. We analyze the skills extracted by multiple different systems and compare those to the skills selected by the annotators, and find that the skills extracted differ a lot in terms of length and semantic content. The skills extracted using conversational large language models (LLMs) tend to be very long and detailed; other systems are concise, whereas humans are in the middle. In terms of semantic similarity, conversational LLMs are closer to human outputs than other systems. Our analysis proposes a different perspective to understand the well-studied, but still unsolved, skill extraction task. Finally, we provide policy recommendations for skill extraction aligned with both HRM and computational perspectives.

The article addresses the challenges that increasing capabilities pose to freedom of thought and conscience. The author contrasts the ongoing debate about the ability to manipulate the human mind with the deeply rooted theories and judicial treatments of freedom of thought and conscience and argues for renewed attention to the subject. Briefly noting that the issue of AI manipulation has especially preoccupied legal scholars, institutions, and societies at large for its impact on the economic market, religious extremism, and politics, the author argues that the capabilities of AI stretch well beyond these limited fields as they challenge the centuries-long notion of freedom of conscience and thought. Because AI capabilities can extract information about individuals’ thoughts and feelings, they are making obsolete the deeply held belief that what is within the human mind (forum internum) is protected from intrusion by its very nature. The author explores the notion of the extended mind to show how the frequent reliance on smart technologies reveals a need to strengthen their protection in order to protect their users. Finally, the author criticizes the widespread focus on the notion of manipulation in recent developments across a range of disciplines and argues for a stronger and more direct consideration of freedom of thought and conscience to update their legal and academic treatment.

This article develops a virtue-epistemological distinction between general and narrow artificial intelligence (AI) systems by analysing AI competence through Ernest Sosa’s virtue-theoretic framework. First, I outline a general distinction between narrow and general AI. I then characterise AI in a way that bridges its terminology with virtue epistemological concepts. Drawing on Sosa’s idea that the telic structure is replicable for systems with merely functional-teleological aims, I argue that current AI systems can possess first-order competence. I argue that increasing domain-specific competence of narrow AI cannot elevate it to the status of general AI, as that requires the constitutional disposition to assess the aptness of one’s own first-order performances, a second-order competence that no amount of first-order improvement can grant. Finally, I argue that the virtue-theoretic distinction between artificial narrow intelligence and artificial general intelligence depends on whether the AI system possesses second-order constitutional competence.

In line with the principle of isomorphism, formally reduced expressions such as clippings or abbreviations develop distinct social and/or semantic functions. This article argues that such differentiation follows the principles of communicative efficiency, with accessibility playing a central role. The claim is supported by a corpus-based comparison of the full form artificial intelligence and the abbreviated form AI in online news and Reddit comments. The short form is used far more frequently in the informal Reddit comments, which rely on a rich common ground, than in the news data. The two forms also display distinct semantic and grammatical profiles. Using distributional semantics (word and sentence embeddings) and Universal Dependencies, I show that the division of semantic and grammatical ‘labor’ between the forms is efficient. The abbreviated form tends to express more accessible meanings related to individual user experience, whereas the full form is more strongly linked to less accessible, abstract meanings of AI as a scientific field, industrial sector, technology or machine capability. In addition, the abbreviated form more often serves as the first element of compounds, whereas the full form often functions as a prepositional modifier, in line with principles of efficient word order.

Artificial intelligence tools for citizen participation have been widely promoted as innovations that can make democratic decision-making more inclusive, efficient, and responsive. Much of the existing debate concentrates on the technical affordances of these tools and the possibilities they create under ideal conditions. While valuable, this focus has obscured a crucial question: who builds, funds, and adopts such tools in practice? We argue that a political economy perspective is necessary to understand the conditions under which AI for citizen participation can meaningfully contribute to democratic governance and how this proposed future may unfold in practice. Drawing on desk research, our experience in relevant research, practice, and policy communities, and informal interviews, we propose a heuristic framework that identifies archetypes of organisations building tools, the funding models that shape their incentives, and the adoption pathways that condition their use. This approach highlights the trade-offs, constraints, and dynamics that influence which tools persist and scale. We suggest that policymakers should not only ask what kinds of tools to adopt but also how to shape an ecosystem that sustains diverse, innovative, and democratically oriented approaches. Our analysis provides an ex-ante framework for situating emerging practices and identifying policy levers to help ensure that AI tools for citizen participation serve the public good.

This article examines how artificial intelligence (AI) impacts state sovereignty and the balance between innovation and control in AI governance through a case study of Türkiye. As AI technologies become increasingly sophisticated, they challenge traditional notions of sovereignty, creating tensions between fostering innovation and maintaining regulatory control. The concept of “AI sovereignty” encompasses a state’s ability to exercise meaningful control over AI infrastructure, data resources, regulatory frameworks, and technological capabilities. Türkiye’s AI strategy illustrates how “middle powers” navigate this balance, especially as it stands at the crossroads of Europe and Asia, requiring Türkiye to develop distinct approaches that reflect its unique geopolitical position. The analysis reveals that sovereignty in the AI domain encompasses multiple dimensions—technical capabilities, regulatory frameworks, and strategic positioning—requiring adaptable governance approaches. The findings offer insights for jurisdictions seeking to balance innovation imperatives with control mechanisms while maintaining strategic autonomy in an evolving global AI landscape.

This article addresses the challenges in formulating data-sharing regulations by proposing a systematic regulatory matrix based on data characteristics. While data-driven innovation drives economic growth, existing legal frameworks in the EU are incoherent and often tend towards data propertisation, even if indirectly, which may lead to data underutilisation. The matrix is built on varied data characteristics, and it aims to foster access to data and data sharing as a fundamental general principle underpinning the data-driven innovation market. This framework offers a balanced regulatory scheme ranging from open access to proprietary models, aiming to maximise innovation and public good in the emerging field of data law.

The paper presents a framework to automatically identify crack patterns and the related features in existing reinforced concrete (RC) bridges. The challenge of this work is to define a tool for detecting the focused defect and highlighting the number and the orientation of cracks, allowing for correct interpretation and driving further evaluations on the residual life of the structure. The study is framed within the increasing interest in monitoring the structural health of existing bridges through automated tools, able to support engineers in the phase of visual assessment and interpretation of structural defects. When dealing with periodic inspection of large bridge portfolios, the support provided by automated tools can be fundamental for planning further strategies aimed at ensuring the structural safety and preventing future disasters. Given a stack of photos of a bridge structural element, an image stitching procedure is proposed to produce a near-complete image of the entire element. On the latter, a pipeline of deep-learning (DL) algorithms is employed to automatically detect and identify cracks (as a combination of object detection and segmentation algorithms). Finally, the proposed tool extracts cracks for counting and defines their orientation (i.e., vertical, horizontal, diagonal), in order to provide near-complete information about the crack pattern for the structural element. A full description of the methodology and the proposed algorithms is reported throughout the manuscript, showing the main pros and cons and assessing the effectiveness of the tool on a real-life case study.

This document presents some advances in personalised and precision nutrition that were addressed in the IUNS-ICN 2025 Congress at August 2025 including the role of biomarkers in personalised and precision nutrition research, personalised and precision nutrition approaches to routine health care services, implications for public health, personalised and precision nutrition around the world, machine learning and precision nutrition, and personalised dietary guidelines alongside population-based policies.

Digital policymaking in the European Union (EU), once seen as an internal market concern, is increasingly shaped by non-economic aims, such as the pursuit of security and the protection of fundamental rights. Recent pieces of legislation, such as the AI Act or the Cyber Resilience Act, have nominally acknowledged the relevance of such factors, but serious concerns have been raised about security considerations de facto trumping all others. In this article, we argue that, despite its predominance, security does not displace fundamental rights or the internal market as the foundations of EU digital law. Instead, we propose a framework to explain how the interaction among rationales for security promotion, rights protection, and market-making goes beyond mere opposition. Applying this framework to three case studies of post-GDPR regulation, we show that the deepening of fundamental rights safeguards in digital regulatory instruments offers, at most, a limited check to creeping securitisation – and sometimes even allows the EU legislator to extend the reach of security measures in the name of protecting certain rights. Understanding the logics and actors that shape the triple helix of markets, rights, and security is therefore crucial for properly understanding – and responding to – security overreach in cyberspace.

The Health Technology Assessment International (HTAi) 2025 annual meeting featured three main plenaries to explore next-generation (NextGen) evidence in health technology assessment (HTA). In this commentary, we capture the discussions of Plenary 2: NextGen Methods: Hype or Here to Stay? Each plenary panelist was tasked to convincingly debate the need, rigor, and implementability of one of three emerging method domains in HTA: (1) Environmental Sustainability, (2) Adaptive HTA, and (3) Artificial Intelligence (AI)-enabled Real-World Evidence (RWE). The three panelists convincingly debated that their method would endure beyond initial hype; all three methods were perceived to have a moderate to high level of need, rigor, and implementability by the audience. Key questions from the audience included a request for examples of where HTA reviews have considered environmental sustainability, a challenge for adaptive HTA to embrace other value elements outside of cost-effectiveness, and a question about how the human-in-the-loop principle fits into AI-driven RWE and what this means for HTA agencies that are already stretched for resources. In this commentary, we summarize the presentations, discussions, and audience engagement to provide readers with accessible insight into the debate about which method(s) are anticipated to endure beyond their initial hype.