ABSTRACT. This paper presents a bibliometric review of the evolving academic literature addressing how exponential technologies (re)shape organizations and work, and their implications for employee outcomes, specifically employee wellbeing and inclusion. The paper adopted a bibliometric review methodology on a sample of 613 papers extracted from the Scopus database, to identify extant re-search on this theme. Publication trends, most cited papers, most frequent words, most productive journals/authors/countries were described in the con-text of bibliometric analysis. Our preliminary bibliometric findings suggest that research on exponential technologies is rapidly expanding and maturing, although it remains fragmented across disciplinary boundaries. China and the United States dominate the field, reflecting their strong investments in techno-logical innovation. Publication growth closely aligns with major technological and societal frameworks, including Industry 4.0, Society 5.0, Industry 5.0, and, more recently, the rise of generative AI. The literature has become in-creasingly AI-centric, with emerging attention to generative AI, algorithmic management, human–robot interaction, and trust, signaling a shift toward so-cio-technical concerns. The emerging prominence of themes such as job inse-curity and human resource management highlights growing interest in the im-plications of exponential technologies for employee wellbeing and workplace inclusion. Overall, the findings illustrate the benefit of adopting a multi-level research agenda linking technological developments to organizational trans-formation and employee experiences.

ABSTRACT. Artificial intelligence (AI) is increasingly used for workplace tasks that rely on employees’ judgement, yet its effects on their cognitive processes (i.e., decision-making and problem-solving) are not well understood. This study examines how AI adoption affects employees’ cognitive processes, and the role critical thinking (CT) plays in these effects. It reports an in-progress systematic literature review (SLR), following the PRISMA protocol, in which 20 of the 206 articles retained for full-text review have been coded so far through a thematic analysis that combines Facione’s six facets of CT with inductive coding. These 20 are the first articles to have completed both coding passes rather than a sample chosen to represent the full set, so the patterns reported here are preliminary. On the evidence coded so far, AI does not appear to have a single effect on cognition. The same tools are associated with stronger cognitive processes in some studies and weaker ones in others, which is consistent with the automation-augmentation paradox. Decision-making is the process most frequently examined, whereas the coded studies do not yet support an answer about problem-solving. CT appears as one condition on the direction of the effect, acting mainly as a moderator and, on thinner evidence, as a mediator. The study therefore proposes a preliminary process model of AI’s effect on cognitive processes and an initial definition of AI-Related Critical Thinking (AI-CT), a construct of four core dimensions and one emerging dimension, both offered as a provisional basis for the completed review.

ABSTRACT. The rapid diffusion of digital technologies and artificial intelligence in workplaces requires an update of the interpretive frameworks used to analyse organ-izational well-being. This systematic literature review addresses the gap concerning the integration of workplace digitalization with the Job Demands-Resources (JD-R) model. It examines how digital technologies, AI and Generative AI (GenAI) operate as job demands and/or job resources, shaping burnout, engagement and workers' well-being. The findings show that digitalization generates specific digital job demands, including technooverload, job insecurity, continuous availability, cognitive demands, algorithmic monitoring and system dependency. At the same time, it may create digital job resources such as job autonomy, digital collaboration, organizational support, career development and efficiency. The interaction between demands and resources is non-linear and depends on buffering and booster effects, the matching between stressors and support, and personal resources such as self-efficacy, digital literacy and learning orientation. AI emerges as an ambivalent factor of work redesign, while GenAI appears as a conditional job resource whose positive effects depend on GenAI self-efficacy, GenAI valuing, human supervision and team cohesion. The review contributes by systematizing recent evidence on digital transformation within the JD-R framework and by proposing integrated conceptual frameworks for digitally sustainable, AI-driven and Generative AI-enabled work environments.

ABSTRACT. This research employed a time series learning algorithm to perform a sustainability analysis on an organization with LSTM neural networks. The main significance of this study is to develop a sustainability model that includes a range of variables, including energy consumption, productivity, and emissions, in a predictive model. Compared to other traditional static models, this sustainability model is much more effective for prediction purposes since it takes into account temporal dependencies along with dynamics through a sequence learning approach. Furthermore, an entire pipeline has been created from data preprocessing to sequence creation, LSTM prediction, and performance evaluation using some standard regression metrics. The study illustrates how Deep Learning (DL) algorithms can assist organizations with data driven decisions in terms of sustainability.

ABSTRACT. Startup success is determined largely by its ability to handle the financial aspect as well as business resource aspects. This research investigates the effect of financial and business factors on startup success using a Machine Learning (ML) approach. Financial and business features data were collected from the Startup Success Prediction dataset, where financial factors include funding volume, funding rounds, and the number of investors. In the prediction model, startups are defined by two categories based on their outcomes based on certain financial features. Startups with acquired outcomes were marked as successful, and closed startups were labeled as not successful. To find out important financial factors contributing to the startup, several ML models, like Random Forest (RF) and Support Vector Machine (SVM), have been applied. Models were tested using the four metrics: Accuracy, Precision, Recall, F1 score, and ROC-AUC. Results show that funding volume, funding rounds, and investor involvement significantly influence the success of a startup. These findings contribute to sustainability-oriented management control systems (MCS) by demonstrating how ML on financial and relational data can inform decisions for venture longevity, which is a key pillar of business sustainability. Predicting venture longevity enables robust economic sustainability assessment, which is a prerequisite for credible long-term ESG reporting.

ABSTRACT. The rapid advancement of artificial intelligence (AI) has transformed work-places across industries, creating both opportunities and challenges for em-ployees. In today's uncertain and dynamic work environment, employees face increasingly complex and diverse work demands. Therefore, the ability to remain resilient has become more important than ever. Resilient employ-ees not only cope effectively with challenging situations but also experience higher levels of well-being and demonstrate better performance. Given the growing importance of employee resilience and the evolving role of AI in the workplace, an important question concerns how these two constructs are re-lated. This paper presents a systematic literature review addressing s the fol-lowing research question: How is employee resilience conceptualized and investigated in the context of AI in the scientific literature? Data were col-lected from the Web of Science database. The keywords used for the search were ("Artificial Intelligence" OR "AI") AND ("employee resilience" OR "re-silience at work" OR "workforce resilience"). The final dataset consisted of 14 publications (2024-2026 period). The findings indicate that research on AI and employee resilience is predominantly grounded in COR and JD-R theories. The analysis also highlighted that employee resilience is both influ-enced by AI and instrumental in shaping how employees respond to AI-related opportunities and challenges in the workplace. By synthesizing the existing knowledge base, this paper provides a comprehensive overview of the field and proposes directions for future research.

ABSTRACT. Global forests are key components to the Earth’s climate system, yet their dynamics exhibit complex statistical behaviors, including multimodal distributions and extreme events, that conventional analytical frameworks fail to adequately capture. This study introduces the Transformed Logistic Symmetric Bivariate Extreme Value (TLSBEV) modelintegrates multimodal marginal distributions with tail dependence structures, enabling simultaneous characterization of both typical forest dynamics and extreme disturbance events. Applied to metrics of tree cover loss, aboveground biomass reduction, and carbon fluxes across 21 years of global data, the model reveals distinct regional signatures: South America demonstrates the highest carbon sink reliability (P(Removals > Emissions) = 0.72), while Southeast Asia shows emerging vulnerabilities (P = 0.58) and Africa exhibits intermediate resilience (P = 0.65). Tail dependence analysis uncovers strong associations between extreme tree cover loss and biomass reduction (λU = 0.45), with carbon flux extremes exhibiting even stronger coupled behavior (λU = 0.52). The TLSBEV framework substantially outperforms traditional bivariate extreme value models (AIC: 1145.2 vs. 1818.7) and Gaussian copula approaches (RMSE: 2.45 vs. 5.67), demonstrating the capacity for capturing both the multimodal nature of forest metrics and the extreme dependencies that characterize large-scale disturbance events.

ABSTRACT. Evapotranspiration is a key component of the hydrological cycle, representing the total flux of water from the earth’s surface into the atmosphere through evaporation and plant transpiration. The methodology involves the extraction of climatological data, including temperature, precipitation, humidity, wind speed, and solar radiation, from various meteorological stations spread throughout Lithuania. The aim was to construct a parsimonious model with explanatory variables exhibiting substantial influence on evapotranspiration, thereby enhancing the model’s interpretability and predictive accuracy. Iterative refinements of the model allowed the optimization of variables that contributed to the prediction of evapotranspiration in the studied region. For all models, the efficiency of the disease prediction models was measured using coefficient of determination (R2) values. The results indicate a significant impact of temperature and solar radiation on evapotranspiration rates in Lithuania. These findings corroborate existing studies that affirm the positive correlation between these climatic variables and evapotranspiration. The constructed stepwise regression model demonstrated proficiency in forecasting evapotranspiration, with satisfactory goodness-of-fit measures and predictive precision.

ABSTRACT. Generative AI (GenAI) promises significant value, yet companies struggle to convert general-purpose GenAI models into reliable commercial prod-ucts. Through an in-depth single case study of an AI-native startup custom-izing a GenAI model for a personalized fitness application, this paper iden-tifies the “GenAI customization paradox” as a fundamental tension where startups must simultaneously leverage GenAI’s open-ended generative ca-pabilities and constrain those same capabilities to achieve viability, com-pliance, and reliability. Drawing on paradox theory and digital generativity, the analysis traces how this paradox manifests across three interrelated di-mensions (i.e., economic, regulatory, and technical), each requiring contin-uous calibration. The study contributes to research on GenAI customization by conceptualizing these challenges as a persistent structural tension, and by identifying the GenAI customization paradox as the mechanism through which the tension propagates. It contributes to digital generativity research by empirically showing that, in GenAI, constraining the generative capabil-ity also impacts the dynamics through which value is created.

ABSTRACT. Psychometric scales are foundational to social sciences, yet their development remains an issue, especially in field contexts where access to respondents is constrained and repeated administrations are infeasible. This is often the case in Organization and Human Resources (HR) research. In response to these challenges, this study explores and evaluates a possible ex ante strategy for improving effectiveness and parsimony in psychometric questionnaire design that leverages large language models (LLMs) to forecast inter-item correlations before data collection. We test this approach on the 2024 U.S. Federal Employee Viewpoint Survey (FEVS) dataset, leveraging its large sample size to obtain stable empirical item-item correlations. Two LLM access modes are compared: (a) embedding-based semantic similarity via cosine similarity, and (b) prompt-based contextual judgments, with prompting explicitly designed to elicit expected empirical correlations. The results show that both approaches capture non-random structure consistent with the empirical correlation matrix. However, based on our results, prompt-based estimates show substantially stronger alignment with empirical correlations than embedding-based estimates, across multiple metrics. We argue that the similarity between prompt-based estimates and real-world empirical correlations resulting from our analysis is sufficiently promising to invite further research on the emerging role of LLMs in psychometric questionnaire design in HR studies.

ABSTRACT. AI-generated visual content is reshaping digital marketing, yet the psychological mechanisms through which it influences consumer behaviour remain underexplored. Drawing on the Stimulus-Organism-Response framework and dual-process theories of persuasion, this study investigates how AI/Human Generated Content Awareness affects Purchase Intention through Perceived Trustworthiness (ability and benevolence) and Clarity. A survey-based experiment (N = 156) was conducted using 15 advertising images for a fictitious fashion brand, randomised across AI-generated and human-generated conditions. Two parallel mediation models were tested, operationalising AI awareness objectively (image type) and subjectively (perceived AI awareness). Results suggest a dual-process dissociation: objective exposure has a positive indirect effect on Purchase Intention via enhanced Benevolence, whereas subjective awareness is associated with a negative indirect effect via lower perceived Clarity. The findings indicate that consumer response to AI content depends on the psychological pathway through which it is processed, with implications for AI-content labelling policies and brand communication strategies.

ABSTRACT. Large language models (LLMs) are increasingly embedded in the conversa-tional environments through which organisations coordinate work. They assist in drafting, summarising, recommending, and reformulating organisa-tional communication. Yet while LLMs improve fluency and efficiency, they also expose a hidden fragility in organisational coordination: conver-sational systems preserve messages more readily than they preserve the so-cio-technical conditions under which commitments become meaningful and binding. This paper argues that organisational commitments do not exist directly in messages, approvals, or workflow states. They emerge through trajectories of interpretation, negotiation, acceptance, and institutional stabilisation. In LLM-mediated conversational work, these trajectories become harder to re-construct because authorship, intention, endorsement, and authority may become partially decoupled from linguistic expression. The result is an in-creased risk of phantom approvals: situations in which conversational trac-es are retrospectively treated as evidence of commitment despite insuffi-cient stabilisation of the underlying communicative trajectory. Building on Language-Action Theory and socio-technical research in In-formation Systems, the paper develops a triadic and processual account of commitment formation. It distinguishes commitment from acknowledge-ment, agreement, approval, authorisation, obligation, and accountability; explains how LLMs destabilise the relation between actors in roles, com-municative acts, and institutional mediation; and presents a process model of commitment as a trajectory rather than an event. The paper further derives design implications for systems supporting con-versational work. It proposes a minimal accountability wrapper as an illus-trative interaction-layer intervention that records selected markers—role, interactional purpose, discourse-act type, and explicit acceptance or re-fusal—without attempting to formalise the full richness of organisational communication.

ABSTRACT. The increasing adoption of artificial intelligence (AI) and digital technologies is reshaping healthcare organizations, transforming not only clinical and admi-nistrative processes but also the ways professionals communicate, collaborate, and experience work. While existing research has largely focused on the tech-nological capabilities of AI, less attention has been devoted to understanding the organizational and relational dynamics through which healthcare professio-nals experience digital transformation. Adopting a human-centered and socio-technical perspective, this study examines how communication practices, leadership support, and organizational alignment shape healthcare professionals’ experiences within AI-enabled work environments. Drawing on the Job Demands–Resources (JD-R) framework, the study employs a qualitative research design based on semi-structured inter-views with 45 healthcare professionals working across different healthcare set-tings. Data were analyzed using thematic analysis. The findings reveal three interrelated themes. First, communication emer-ged as a paradoxical organizational mechanism, functioning both as a resource that facilitates collaboration and as a demand that contributes to information overload. Second, leadership was identified as a key organizational resource, with support, recognition, and participation facilitating adaptation to technolo-gical change. Third, organizational alignment influenced how AI initiatives were interpreted, accepted, and integrated into everyday professional practice. The study contributes to the literature on AI-enabled organizing by showing that AI is not experienced as an inherently beneficial technology. Rather, its effects are mediated by organizational conditions and socio-technical dyna-mics. Communication, leadership, and organizational alignment shape whether AI-enabled systems are perceived as meaningful resources or as additional or-ganizational demands. The findings highlight the importance of human-centered approaches to digital transformation in healthcare.

ABSTRACT. Healthcare handoffs, the moments in which responsibility for patients is transferred between clinicians, are conventionally framed as episodes of information transfer whose failures can be remedied through standardized communication protocols. This framing is increasingly inadequate for understanding handoffs mediated by artificial intelligence (AI). Drawing on Orlikowski’s sociomateriality and technology-in-use perspective, complemented by the concepts of affordances and interpretive flexibility, this paper reconceptualizes handoffs as situated sociomaterial practices in which clinical judgment, accountability, handwritten notes, electronic health records, temporal pressures, and professional expertise are constitutively entangled. We argue that AI does not simply support, automate, or optimize handoffs: it may reconfigure what becomes visible, what is prioritized, what is documented and remembered, and how responsibility is distributed across clinicians and technological artefacts. The argument is illustrated through a preliminary qualitative study of an internal medicine department in a Northern Italian hospital. The paper contributes by extending sociomaterial theorizing to AI-mediated clinical coordination and by providing a framework for analyzing the affordances, performative effects, and contested meanings of AI in safety-critical organizational routines.

ABSTRACT. Digital Transformation (DT) confronts organizations with complex socio-technical challenges that encompass technological innovation, data management, governance, and strategic decision-making. Intermediary organizations, such as employer associations, are particularly affected due to their knowledge-intensive operations and their role as coordination hubs within business ecosystems. This paper presents the empirical validation of S.Me.Co. (Shared Memory Collective), a Generative AI–based cognitive infrastructure designed to support strategic decision-making in intermediary organizations. The system integrates tacit knowledge, structured documents, and adaptive Retrieval-Augmented Generation (RAG) mechanisms to operationalize advanced data management principles. By transforming unformalized information into a continuously evolving organizational memory, S.Me.Co. ensures information traceability, coherence, and controlled access across institutional and firm-level interactions. A mixed-method research design was adopted, combining a qualitative pilot phase with a large-scale quantitative evaluation involving 995 inquiries from 308 firms. To assess system performance, the study introduces a multidimensional Impact Assessment Framework (IAF), which evaluates response coherence, feedback utilization, and user engagement. Results indicate a substantial improvement in information coherence (from 84% to 95%) and a 49% reduction in response time, corresponding to 1.272 hours saved. Furthermore, the system enhances decision-support capabilities and reduces information asymmetries within the intermediary ecosystem. The findings demonstrate the potential of Generative AI–enabled cognitive infrastructures to improve performance analytics, support data-driven governance, and enable scalable digital transformation in complex organizational networks, with specific reference to the national Confindustria system.

ABSTRACT. The key challenge with conventional Geographical Information Systems (GIS) lies in their static nature: maps are on-demand snapshots of a territory rather than dynamic, interactive tools. Business Intelligence (BI), by contrast, requires interactivity, with maps generated within seconds and insights extracted from maps in real time. We have developed an innovative dashboard, called Gen4Geo, that bridges this gap, supporting the real-time exploration of large geospatial datasets in natural language. In this paper, we report on the results of the application of Gen4Geo to the fragility assessment use case in order to 1) prove the technical feasibility of the approach, 2) understand whether a BI approach is perceived as useful by a community that is tradi-tionally active with GIS, 3) collect preliminary evidence on the extent to which AI can reduce the time needed by analysts to discover insights, and 4) gather preliminary feedback on the ease of use of the platform by non-technical end users. In this paper, we present our technology platform, de-scribe our methodology for the assessment of fragility at subnational level, and report on its application to Mozambique. Our preliminary findings sug-gest that Gen4Geo is technically feasible and that analysts were able to ex-tract geospatial insights significantly faster than with traditional GIS work-flows. Initial observations from non-technical users indicate that the natu-ral language interface lowers the barrier to geospatial analysis, making the platform accessible beyond the GIS specialist community.

ABSTRACT. This study aims to understand how scientific literature examined the connection between AI, technologies that use AI directly and/or indirectly, and sustainable tourism through a Literature Review and Bibliometric analysis using the database Scopus. A dataset of 165 publications was analyzed through citation, co-citation and co-occurrence analysis. Results show a regular trend in publications and an irregular trend in citations over the years and identify the most active sources. Co-citation analysis shows a network of 5 clusters with 110 items and 1476 links, clusters to which we then associated the papers emerged in Scopus after having identified their references and their correspondence with the papers identified in VoSviewer, placing each in the most suitable thematic cluster, association confirmed through a careful reading of them. Co-occurrence analysis shows a network of 7 clusters with 62 items and 522 links identifying specific thematic clusters, relating the topic, and thematic connections between them. Discussion of Literature Review shows the type of connection between AI technologies, others technologies that use AI, and sustainable tourism; technologies most considered and specific types of sustainable tourism considered in the literature. Scholars could use this study as an initial framework to understand the state of the art of contributions on the topic in literature, useful to direct future analyses. Policy makers could benefit from systematically organizing existing knowledge on the topic to better understand how to direct national and international policies toward more effective management of new technologies, like AI, for more responsible and sustainable tourism.

ABSTRACT. Artificial intelligence (AI) is widely framed as a technology of efficiency and organizational optimization. This paper challenges this view by conceptualizing AI adoption as a recursive socio-technical metabolism in which organizational processes transform AI-enabled efficiency gains into expan-sionary trajectories across infrastructure, labor, and knowledge systems. Moving beyond fragmented accounts of environmental, economic, and social externalities, we argue that AI-related unsustainability does not arise from isolated effects, but from a unified expansionary dynamic embedded in the operational logic of AI adoption. Organizational adoption and integration processes transform AI-enabled efficiency gains into system-wide increases in computational demand, infrastructural dependence, and organizational complexity. Efficiency gains do not automatically reduce overall resource consumption; instead, by lowering marginal costs, they often enable broader adoption and socio-technical expansion, potentially intensifying rather than alleviating the constraints they are intended to address. This logic extends to organizational cognition and epistemic infrastructures. As AI is embedded in decision-making, organizations increasingly rely on opaque and probabilistic knowledge production, reshaping how uncertainty is constructed and governed. We theorize AI as an infrastructural and organizational regime characterized by recursive feedback between optimization, scaling, and externalization. Governance is not external to this system but emerges within it, simultaneously enabling and stabilizing expansion. Sustainability is not an exogenous objective achievable through better management, but a property destabilized by optimization itself. We conclude that the sustainability implications of AI adoption cannot be addressed within efficiency-centric paradigms, as efficiency is the primary enabling condition of systemic growth, calling for constraint-based AI governance.

ABSTRACT. Household waste electrical and electronic equipment (WEEE) is a demanding case for public-service digitalization because a household disposal decision must pass through service, organizational and material arrangements before it becomes an authorized return. The study asks under which conditions digi-talization can support this transition. It combines an integrative literature re-view with diagnostic bibliometric analysis of 7,334 Scopus and Web of Sci-ence records. Five query families represent three complementary streams: household WEEE governance, digital government and citizen-facing public services, and Digital Infrastructure Theory. Thematic mapping and biblio-graphic coupling reveal selective intellectual proximity but no integrated knowledge base spanning all three streams. A common eligibility protocol retained 47 primary and 11 supporting deep-evidence contributions for cross-stream synthesis. Mechanisms were compared across streams by the frictions they address and the level at which they operate. The analysis iden-tifies four second-order conditions: citizen access and participation, citizen-service interface quality, public digital coordination capacity, and material return convenience. Cross-level infrastructure alignment captures compatibil-ity among these conditions rather than adding a fifth. The framework redi-rects evaluation from digital adoption to the configuration linking citizen en-gagement with verified physical return. Its five propositions remain untested and require context-sensitive empirical validation.

ABSTRACT. Food technology (so-called food tech) has emerged as one of the most dynamic and strategically important phenomena in the modern global economy, but its debatein the business and management domains is fragmented . In order to bridge up this gap, this study conducted a bibliometric analysis of 228 papers that were published between 1985 and 2026. These papers were obtained from Scopus using a methodical query that included science mapping and performance analysis. We tracked the conceptual development of food technology research in business and management using the Bibliometrix package in R, finding publication patterns, the most productive authors and sources, and the field's thematic structure. According to our research, academic production increased exponentially starting in 2018, with consumer acceptance, agri-food systems, and sustainability as arising key themes. Five thematic clusters are identified by the scientific mapping study, which highlights both consolidated research streams and intriguing prospects for future exploration. These clusters include developing sectors like food tech entrepreneurship and digital platforms in emerging countries. This study depicts the state of the art of the current and strategic themes in this novel field, providing theoretical and managerial contributions, useful for the policy-makers, also.

ABSTRACT. This study seeks to explore socio-technical barriers for integrating artificial intelligence-based solutions with lean manufacturing principles in cheese manufacturing small and medium-sized enterprises characterized by high variability and tacit knowledge․ Although there is a rapidly growing interest in combining artificial intelligence with lean manufacturing‚ the factors that make implementation difficult in human-centric and highly variable production environments remain underexplored. The qualitative research design‚ based on ten semi-structured interviews with managers involved in lean and artificial intelligence projects and subsequent data analysis based on the Gioia methodology‚ highlights the idea that integration challenges between lean and artificial intelligence projects are not merely technical challenges‚ but rather systemic misalignments associated with social-technical systems theory․ Four dimensions of these barriers are identified, which are interrelated with each other: organizational readiness; strategic & economic misalignment; data & process readiness; and technical infrastructure constraints. The findings show how artificial intelligence disrupts tacit knowledge-based lean routines‚ changes decision-making modes‚ and raises trust‚ data‚ and operational stability dilemmas․ Based on these findings‚ a joint optimization framework is proposed to improve capability building, participatory change, data standardization, and modular technological integration․

ABSTRACT. . The tourism industry and destination management field face increasing pressure to develop amid rising demand for digital and sustainable manage-ment solutions. This study examines how Lean Management (LM) principles, known for reducing variability and enhancing quality through continuous improvement, can be further sustained by developing a digital platform to increase social sustainability attention and service excellence achievement in the tourism field and destination management area. While LSS frameworks offer structured tools for process optimization, their effectiveness in tourism is hindered by poor data standardization and low inter-organizational trust. We propose a digital platform designed model which facilitates data organi-zation for complex economic ecosystems such as a tourist destination and increases the focus on human centricity and social sustainability for workers and all stakeholders involved within the service construction and delivery process. The model facilitates real-time transparency and stakeholder ac-countability, This new digital platform called ImpactSync aims to rationalize the stake-holders relationship management and organization focusing on the use of Lean Management principles within the variable and fragmented context of tourism. While theoretical, the model is based on the lean principles solid and tested through the years in many different sectors, the digital platform design opens space for future empirical testing and refinement. It contributes to emerging literature on digitally enabled service ecosystems and highlights the synergistic potential of combining LSS methodologies in destination management field, and service industry aiming for social sustainable practic-es implementation and service excellence delivery

ABSTRACT. Organizations are increasingly required to respond to the challenges of digi-tal transformation to remain competitive over time. This requires them to redesign their internal processes considering the widespread diffusion of digital technologies. Public administrations are not immune to these pres-sures for change and are likewise required to address the challenges asso-ciated with digital transformation. In the public sector, however, digital transformation does not simply respond to citizens’ and users’ growing ex-pectations for accessibility, efficiency, and service quality. It also assumes an institutional meaning. For this reason, digital transformation in public administration should not be understood merely as the reorganization of in-ternal processes, but rather as a broader process of institutional and organ-izational change. This study adopts a qualitative, descriptive-exploratory approach based on a single case study. The empirical case concerns a pub-lic administration that redesigned its revenue collection process between 2023 and 2025. The analysis is based on internal reports and procedural manuals, which allowed us to reconstruct both the historical AS-IS configu-ration and the redesigned TO-BE configuration. The AS-IS/TO-BE compari-son is conducted through a set of analytical measures aimed at examining how digital transformation reshapes organizational routines, information flows, time requirements, and coordination mechanisms. The findings show that digital transformation generates organizational value when it is accompanied by the redesign of public service processes. The study offers theoretical and practical insights for public managers seeking to initiate digital change and transformation processes through process redesign.

ABSTRACT. The twin transition—the convergence of digital transformation and sustainability has become a central objective in European policy and industrial agendas. While artificial intelligence (AI) is increasingly portrayed as a key enabler of both competitiveness and sustainability, SMEs interpret and enact these pressures in highly heterogeneous ways. Drawing on sensemaking theory and institutional perspectives, this study examines how SMEs construct meaning around AI within the broader twin transition agenda. We analyze qualitative data collected within a regional initiative involving 184 SMEs, including 19 semi-structured interviews, participant observations, and field notes. Using a Gioia-inspired methodology, we identify four recurring organizational responses: non-interpretation, selective compliance, experimental adaptation, and institutional integration. These responses represent distinct sensemaking processes through which firms assess the legitimacy, relevance, and feasibility of twin transition initiatives. Our findings show that institutional pressures do not produce uniform adoption outcomes. Instead, SMEs actively interpret, negotiate, and selectively enact competing expectations, leading to distinct strategic trajectories. By shifting attention from adoption outcomes to interpretive processes, this study contributes to information systems research on digital transformation and the twin transition by offering a processual explanation of how SMEs construct meaning around AI-enabled transformation within the broader twin transition agenda.

ABSTRACT. Algorithmic systems increasingly govern HR decisions in allocation, evaluation, scheduling, and reward, applying neutral rules to workers. We ask whether such procedurally neutral governance can produce occupationally stratified decision latitude, and which arrangements reduce, or widen, the gap. The central issue is not whether a system is contestable by design but whether contestability can be exercised equally by differently positioned workers. We reconceptualize contestability as an occupationally distributed organizational capability and test it with an agent-based model in which three occupational groups face the same algorithmic rule but differ in four dimensions of contestability—contestation cost, interpretive access, voice weight, and outside-option dependency—linked by a standing-related feedback loop. Under homogeneity the rule produces no stratification, whereas, under occupationally patterned heterogeneity a relevant gap emerges, opened by the static distribution of contestability and amplified by the feedback loop. Two governance findings follow. First, uniform transparency narrows the gap where interpretive access binds and widens it where voice binds. Second, an independent appeal, moving the human-in-the-loop boundary beyond line management, widens the gap when upstream capacities to detect and attempt contestation remain stratified, and compresses it only when those capacities are equalized. A 2⁴ decomposition shows that interpretive support and voice equalization deliver almost all of the bundle's compression, more than halving the gap, while override floors and standing protection do little alone. Contestability mechanisms are double-edged unless capacity is equalized, therefore, their distribution should be the object of governance.

ABSTRACT. AI technologies are developing at a rapid pace these days and require more infrastructure deployment. As we write this paper, we see new data center construction initiatives being discussed daily, but we also observe debates about introducing regulatory controls on data center placement. Data centers have both positive and negative aspects. While the scientific and managerial literature primarily focuses on the environmental and economic impacts, the social aspect of this phenomenon remains largely overlooked. In this paper, we would like to contribute to the social discourse on data center deployment and examine different approaches to data center infrastructure development using the example of two countries, Ireland and the United Kingdom, through the lens of the S-LCA approach. The results showed that, despite growing attention to the environmental and economic implications of data center development, policy documents still provide limited consideration of potential social benefits and social costs living room for more comprehen-sive social impact assessment.

ABSTRACT. Universities are increasingly required to balance competitive positioning with sustainability objectives while operating under conditions of institutional complexity, stakeholder pressure, and resource constraints. In this context, strategic management and corporate social responsibility are often treated as distinct domains, operationalized through the Strategic Plan and the Sustainability Report. However, these tools rely on a largely shared informational base and are more interconnected than typically assumed. This paper argues that their relationship should be understood as an integrated governance cycle rather than as a linear process. The central claim is that Artificial Intelligence (AI) enables this integration by acting as a cognitive and governance infrastructure. AI enhances organizations’ capacity to collect, process, and interpret heterogeneous data, supporting decision-making, organizational learning, and performance evaluation (Krakowski et al., 2023). Within this framework, AI performs four key functions: integrating data across planning and reporting systems, supporting strategic sense-making by identifying gaps between expected and observed outcomes, enhancing accountability through more timely and data-driven reporting, and providing predictive and prescriptive insights for adaptive strategy. As a result, the Sustainability Report can be reinterpreted not only as a disclosure tool, but also as a strategic measurement and feedback mechanism that informs decision-making processes (Laine, 2024). The proposed framework highlights the role of AI in connecting planning, action, and reporting within a unified governance system, contributing to the literature by reframing the relationship between strategy and sustainability.

ABSTRACT. Genomic surveillance networks built during the COVID-19 pandemic must now sustain laboratory participation as emergency funding and attention recede. Coordinators of such infrastructures need early visibility into whether individual laboratories can sustain the specific work the network depends on. Existing assessment tools primarily address national preparedness or the sustainability of laboratories as standalone organizations. This study asks how the sustainability of laboratory participation can be diagnosed at the node level in an auditable, comparable, and actionable form. Through Action Design Research in the Italian network for genomic surveillance of SARS-CoV-2 and related variants, we design the Network Participation Sustainability Index (NPSI), a formative composite index comprising five dimensions: infrastructure resilience, human capital, operational performance, operational robustness, and institutional environment. We report the first design iteration and a pilot application using survey data from five laboratories. The pilot demonstrates that the scoring architecture can be applied to real data and reveals materially different dimensional profiles behind relatively similar composite scores. The study contributes four provisional design principles for diagnosing the sustainability of node-level participation in similar coordinated health information infrastructures.

ABSTRACT. Public-health and environmental open data are abundant, but the two streams rarely travel together. Mortality registries and pollution indices are published by separate agencies, on separate portals, with separate ontologies and rationales; the analyst who wants to read them jointly has to do the merging herself, in a notebook that rarely returns to a decision-maker. We present a dashboard that closes part of this gap on Californian open data. The artifact joins the state’s mortality file by ZIP code with the CalEnviroScreen 4.0 environmental index, and then interrogates the merged surface through three deliberately heterogeneous lenses: Local Moran’s I for spatial cluster detection, Random Forest permutation importance for predictor ranking per cause of death, and Pearson and Spearman correlation with effect-size and significance bands for bivariate strength. None of the three methods, on its own, supports a defensible claim about an environment–mortality association. Their triangulation does: a hotspot that survives all three filters is a candidate for confirmatory work, while disagreement among the lenses is itself diagnostic. An Italian replication reinforces the argument: when Random Forest assigns top importance to socioeconomic indicators acting as spatial proxies for industrial pollution, the bivariate lens recovers the direct particulate-matter–cancer association that the multivariate ranking had distributed across correlated predictors. We frame the artifact as a boundary object in the sense of Star and Griesemer, argue that the multi-lens architecture is what gives the dashboard cross-community readability, and reflect on transferability to other national health systems built on comparable open-data infrastructures.

ABSTRACT. Healthcare professionals face alarmingly high rates of workplace vio-lence, highlighting the need for training in non-technical skills to manage challenging patient behavior. Immersive virtual reality has emerged as a promising training modality for this purpose, yet its systematic imple-mentation in healthcare remains limited. This study presents the first two echelons of an echeloned Design Science Research project addressing how immersive virtual reality training can be designed to support non-technical skills development. We conducted semi-structured interviews with eight domain experts from Norwegian healthcare institutions and universities, analyzed using reflexive thematic analysis. Analysis re-vealed sociotechnical misalignment across three problem areas: the in-tegration of immersive virtual reality into training practice, its didactic use in training sessions, and the quality of training materials. Building on these findings and guided by sociotechnical design principles, we de-rived and validated design objectives and requirements for a knowledge platform that supports the systematic design and conduct of immersive virtual reality-based training in healthcare. The knowledge platform pro-vides structured guidance for curriculum integration, facilitation, and material quality assurance, giving practitioners actionable support for implementing immersive virtual reality-based training effectively.

ABSTRACT. This study investigates how physicians’ professional identity influences their voluntary use of health information systems (HIS) in direct patient care and how perceived patient and colleague norms influence the visible enactment of this use. We argue that HIS use in clinical encounters is not merely a tech-nology adoption decision, but also a competence performance: physicians must access information effectively while managing how this information-seeking behavior is interpreted by patients and colleagues. Drawing on 16 semi-structured interviews with physicians from 12 specialties, the study em-ploys an abductive qualitative approach. Four physician identity types emerge from the data – Traditionalist, Authority-Protector, Transparent Em-path, and Pragmatist – each associated with HIS usage patterns ranging from open use, to concealed use, to avoidance and utilitarian use. Our findings show that professional identity shapes whether HIS use is interpreted as competence-enhancing, competence-threatening, transparency-supporting, or pragmatically useful. Perceived social norms amplify, attenuate, or redi-rect these identity-based tendencies depending on whether patient and col-league expectations are experienced as aligned or misaligned with physi-cians’ self-understanding. Contextual factors such as urgency, information sensitivity, and HIS accessibility further shape usage, sometimes overriding both identity and norms. We develop a conceptual model of identity-based HIS use that explains how physicians arrive at different forms of HIS use in socially observable clinical situations. The study extends technology adop-tion research by shifting attention from adoption to enactment and from sys-tem use to the social meaning of use.

ABSTRACT. Emerging technologies, such as generative artificial intelligence, Web 4.0 and virtual worlds, are redefining organisational, educational and institutional processes, particularly in contexts shaped by digital transformation. Immersive environments offer new opportunities for training, collaboration and the development of digital competences, although their integration into public organisations remains a technological challenge. This research was conducted within the context of the European UPRAISE project and presents the design and development process of a virtual environment (https://framevr.io/uninamaster) associated with the Executive MPA’s degree “Trasformazione digitale della Pubblica Amministrazione”, delivered at the University of Naples Federico II. Through a qualitative descriptive methodology, the study examines the planning, construction and implementation process of the virtual environment, involving public-sector practicioners, teaching staff and experts in scientific communication. The environment, developed with Frame VR, is structured into interconnected spaces aimed at promoting the master’s degree, supporting teacher-student interaction and providing access to academic resources. The results suggest that students have limited initial knowledge of virtual worlds, but also a high level of interest in their application within the field of public administration, particularly in training, institutional communication and citizen-administration interaction. It is concluded that virtual worlds can contribute to more active and adaptive training experiences, aligned with contemporary processes of public innovation and digitalisation.

ABSTRACT. The integration of artificial intelligence into operational infrastructures is transforming how organizations generate and enact decisions. In energy sy-stems, AI-based systems monitor grid conditions and generate operational recommendations across distributed network nodes. As electricity net-works evolve toward distributed architectures — conceptualized as an Energy Physical Internet — decision intelligence becomes distributed across infrastructures, creating new governance challenges. While prior re-search has examined behavioral responses to algorithmic systems, conside-rably less attention has been paid to the governance structures that determi-ne how decision authority is allocated between human actors and algorith-mic systems. This paper introduces the concept of authority ambiguity, de-fined as the structural governance condition in which decision rights wi-thin a human–AI decision system are insufficiently specified, leading ac-tors to experience uncertainty regarding whether to exercise, defer, or over-ride algorithmic recommendations. Drawing on IS governance theory, orga-nizational design, and socio-technical infrastructure research, and develo-ped through integrative conceptual theorizing, we propose a conceptual model linking authority ambiguity to three governance failure patterns: escalation paralysis, accountability diffusion, and override reluctance. The domain relevance of the framework is illustrated through the officially in-vestigated Iberian Peninsula blackout of 28 April 2025. The study contri-butes to IS research by (1) introducing authority ambiguity as a structural governance construct, (2) extending decision architecture as a governance mechanism for structuring human–AI decision authority, and (3) explaining how authority ambiguity generates governance failures in distributed digi-tal infrastructures, while specifying structural, cultural, and temporal boun-dary conditions. Implications for energy system operators and regulators are discussed.

ABSTRACT. Organizational culture and help-seeking have both been extensively studied, yet their relationship has been treated almost exclusively as unidirec- tional, with culture shaping the conditions under which employees seek or avoid help. The adoption of hybrid work arrangements has intensified this relationship by making help-seeking more deliberate, more visible, and more consequential as a cultural act. This paper challenges that assumption. Drawing on practice the- ory, complexity theory, and socio-technical systems theory, we argue that the relationship is bidirectional and recursive: culture shapes help-seeking, and help- seeking practices, accumulating and stabilizing over time, actively participate in producing and transforming organizational culture from the bottom up. To ex- plore this dynamic empirically, we draw on 13 semi-structured interviews with HR managers from Italian organizations operating in hybrid work environments. Using inductive thematic analysis, the findings reveal five cultural mechanisms through which culture shapes help-seeking (trust architectures, communication hierarchies, onboarding practices, destigmatization of vulnerability, and environ- mental and symbolic signals) and three processes through which help-seeking practices produce culture (bottom-up policy revision driven by collective re- quests for support, deliberate design of new support rituals to compensate for informal interaction lost in hybrid settings, and cultural erosion caused by sys- tematic non-help-seeking in fragmented organizational structures). This study contributes to the literatures on organizational culture, help-seeking behavior, and hybrid work by offering preliminary empirical evidence for a bidirectional model and by identifying the specific organizational mechanisms through which each direction operates.

ABSTRACT. This paper examines knowledge retention (KR) in manufacturing small and medium-sized enterprises (SMEs) through two complementary digital technologies: Enterprise Resource Planning (ERP) systems and Generative Artificial Intelligence (GenAI). Drawing on an integrated qualitative pool of empirical evidence collected in Italian manufacturing SMEs, the article develops a comparative framework linking digital KR conditions to workers' professional identity. The framework relates three interdependent dimensions - knowledge sharing propensity, technology adaptability for knowledge sharing, and organizational culture - to identity outcomes. Despite their different technological logics, ERP and GenAI generate comparable organizational tensions: both support KR by transforming tacit individual knowledge into more explicit and shared organizational knowledge, while also challenging workers' perceived centrality, control, and recognition. The proposed 3-axis framework offers managers and researchers a diagnostic tool for anticipating identity-related dynamics in technology-mediated KR transformations.

ABSTRACT. In hybrid work settings, perceived decision-making involvement may not carry the same implications for supervisory evaluations as in fully on-site work. This study examines whether top-down communication quality is indi-rectly associated with perceived supervisory quality via employees’ per-ceived decision-making involvement, and whether this association differs across work arrangements (on-site vs. hybrid). Survey data were collected from 566 employees at an Italian public university; structural models were estimated on 553 complete cases. Results show that high-quality top-down communication is positively associated with perceived decision-making in-volvement, which, in turn, is associated with more positive supervisory eval-uations. However, this indirect association is weaker and non-significant among hybrid employees. Drawing on Social Information Processing Theory and Leader–Member Exchange theory, we identify work arrangement as a boundary condition of the association between perceived decision-making involvement and supervisory quality. These findings contribute to research on hybrid work by showing that employee involvement does not carry the same implications for supervisory evaluations across work arrangements.

ABSTRACT. Measurement in the social sciences and in Information Systems (IS) research has traditionally relied on multi-item scales to enhance reliability and validity. However, this approach assumes consistent respondent engagement, an assumption that may not hold in contemporary survey contexts. We introduce the concept of measurement-induced disengagement, defined as the cognitive and affective disengagement triggered by the structure of measurement instruments. Across two studies, we show that single-item measures are strongly correlated with multi-item scales, while multi-item formats increase perceived effort, reduce attention, and generate negative affective responses. These findings highlight a trade-off between psychometric precision and respondent engagement, suggesting that shorter measures may, in some contexts, improve overall measurement quality.

ABSTRACT. Generative AI (GenAI) tools have entered software development workflows at remarkable speed, and they are now woven directly into the practices through which teams coordinate expertise: knowing where expertise sits, recognising where it is needed, and bringing it to bear. Drawing on an inter-pretive study of eight semi-structured interviews with developers and pro-ject managers in Sweden, and reading the material through Faraj and Sproull’s expertise coordination framework, transactive memory systems, and an enacted conception of affordances, we examine what these tools af-ford and constrain in practice. We find that GenAI tools afford immediate and vividly felt individual benefits, automating routine work, filling knowledge gaps, and lowering the barrier to technical problem-solving, while simultaneously constraining the human interactions through which a team’s tacit map of who knows what is built and maintained. We argue that affordance and constraint here are not separate phenomena but two faces of a single structural mismatch: GenAI generates outputs on a statistical sub-strate that is categorically different from the tacit, relational substrate on which expertise coordination depends. The benefits are visible and the costs are tacit, and because both arise from the same action, nothing raises the alarm.