Generative AI (GenAI), particularly large language models (LLMs), is increasingly embedded in decision-making processes, making it essential to understand the drivers of its actual use. Despite their accessibility, LLMs are prone to hallucinations and bias, which contribute to user skepticism. This study examines the factors shaping perceptions of LLMs in personal and professional decision-making contexts, focusing on perceived usefulness, user trust, and personal values. An anonymous survey of 215 professionals was analysed using descriptive statistics, hierarchical cluster analysis, and regression analysis. Findings reveal that, contrary to expectations, trust and perceived usefulness alone do not sufficiently explain LLM adoption. Instead, personal values emerge as a significant yet underexplored determinant influencing usage decisions. These results extend current GenAI adoption research by incorporating the role of individual values. The study highlights the importance of context-sensitive, stakeholder-aware strategies and underscores the need for further research on value alignment across diverse cultural and regulatory environments.
This paper provides a review of DevOps literature published between 2022 and 2026 a significant time period that coincides with the rapid diffusion of generative AI (GenAI) tools in software development practices. We examine over 1060 peer-reviewed papers to categorize the literature through the PPTD (People, Process, Technology, and Data) socio-technical lens. Our work provides a synthesis of dominant trends and themes in the DevOps conversation and how this has changed over time following the emergence of GenAI. A significant increase in research on 'Data' focused topics and GenAI in particular is observed, as well as the continuing paucity of research on 'People' focused topics. Following our analysis, we propose a future research direction: applying an Activity Theory lens to deliver greater insights into the impact of GenAI on the DevOps workflow.
TThis paper proposes the premise of a theory of human-AI collaboration in decision making through an analytic autoethnographic case study. Drawing on a 12-month interaction with a large language model and a mechanical repair episode, the paper supports and extends Simon's work: problem structuredness is not an intrinsic property of a task but depends on the decision maker's representation. The analysis reveals how problem representations are formed, stabilised and revised through interaction. The findings highlight three important mechanisms in human-AI interaction that bear on the evolution of the decision maker's problem representation: hypothesis generation, management of the timing and reversibility of the intervention (labelled 'decision pacing' thereafter) and human override. The study contributes to decision theory by reframing structuredness as epistemic rather than ontological, illustrating the importance problem representation in the process of decision making and showing how AI can stabilise, but also distort, decision makers' problem representations.
The purpose of this study was to conduct a review of the literature in information systems to gain insights into (1) the existing knowledge on website and mobile accessibility across various fields and (2) the methodologies currently employed to assess the willingness of individuals with visual impairments to utilise different technologies. The study collected a total of 18 articles related to websites and mobile applications reviewing different accessibility factors based on the Web Content Accessibility Guidelines (WCAG) 2.2 mobile guidelines. The findings of the study revealed that existing literature focused only on levels A (54.5%) and AA (45.4%) of the WCAG 2.2 mobile success criteria without considering level AAA. Also, half of the collected studies (50%) used quantitative methods and (33.3%) of the collected studies used qualitative methods. The contribution of this study is the identification of research gaps on digital accessibility for people who have visual impairments.
This paper develops and evaluates a human-centered decision support framework for explainable social media message prioritization. Adopting a Design Science Research (DSR) perspective, it addresses message handling as an ordered multicriteria sorting problem. The proposed approach is operationalized through a two-phase method comprising criteria elicitation, reference-message evaluation, rule induction, iterative validation with the decision-maker and application of the validated rule base to new messages. The framework is applied within a French business school for institutional communication on X. Evaluation combines decision-maker validation, dominance-based consistency analysis and stratified 5-fold cross-validation. The results show that the induced rule base provides an interpretable and operational representation of the decision-maker's prioritization logic, with a global quality of approximation of 0.86, a mean accuracy of 0.74 and a balanced accuracy of 0.76.
This paper presents the results of an experiment about humans using a large language model (LLM), analyzing the extent to which humans anchor on numeric rankings provided by an LLM, the extent to which they adjust those rankings and the extent to which those adjustments are 'systematic'. The paper finds that users appear to anchor on those LLM numeric rankings, but users appear to adjust either by decreasing high rankings or increasing low rankings - few took the LLM ranking exactly, in contrast to expectations based on the notion of 'least effort'. This suggests a 'semi-autonomous' bias and hybrid intelligence, whereby people's judgments are affected by the anchor bias. They make a change in the LLM information to generate their own ranking. Although a self-assessment of their 'knowledge' was not rated highly, that knowledge was positively and statistically significantly related to their rankings.
With the development of artificial intelligence, human-machine value co-creation can increasingly contribute to business performance. However, prior research on the interactions between technological transformation and human aspects of corporate decision making is surprisingly limited. By examining whether intuitive or analytic features of managerial decision making style moderate the link between technological transformation (including AI applications) and business performance, our paper aims to contribute to filling this gap. To explore the research questions, data from 335 Hungarian medium-sized and large firms are analysed with regression modelling. Empirical results suggest that, as opposed to a nonsignificant intuitive decision style effect, analytic decision style has a significantly positive moderating effect in the link between technological transformation level and business performance. These results add to previous literature on business-related technology development and also highlight the uniqueness of human intuition in corporate decision making.
Generative artificial intelligence (GenAI) is increasingly embedded in interactive digital systems, yet user adoption remains uneven because users evaluate it as both useful and uncertain. This study examines GenAI adoption as a human-computer interaction process shaped by cognitive evaluations and affective responses in academic and knowledge-intensive settings. Using survey data from 645 respondents, mainly undergraduate and postgraduate users in higher-education contexts, the study applies Partial Least Squares Structural Equation Modeling, SHAP-based machine learning, and fuzzy-set Qualitative Comparative Analysis. The findings show that attitude toward use and perceived control are the strongest proximal drivers of adoption intention. Trust and perceived AI competency influence intention mainly through attitude and perceived control, while anxiety shows a small positive direct association that should be interpreted cautiously. The study contributes by explaining mediation patterns, contradictory effects, and configurational heterogeneity in GenAI adoption.
This qualitative study explores how young people in India sustain difficult decisions in challenging social and personal contexts. Drawing on in-depth interviews with 15 participants, the study examines the role of anam cara ('soul-friend') relationships in supporting emotional resilience and decision sustainability. Four key relational roles emerged: Emotional Witness, Moral Anchor, Social Translator, and Future Mirror. These roles enable individuals to move beyond the limitations of linear Rational Choice models - which often result in overthinking - towards values-aligned decision-making within a Circular Model of Choice grounded in the Logic of Appropriateness. The findings suggest that anam cara relationships function as psychological anchors that provide emotional validation, identity reinforcement, and sustained decision support in moments of uncertainty. This study contributes to the literature by proposing a relational framework that integrates identity, emotional support and decision-making. These findings highlight the importance of relational support systems in fostering emotionally sustainable decision-making among young people.
Limited research has examined how ports' big data analytics capability (BDAC) is associated with operational and sustainable performance. In response, this study develops and validates a decision support system (DSS) that integrates expert judgements, fuzzy set theory, unsupervised machine learning (ML), Decision Trees, and Bayesian Network analysis. Data were collected through a Likert-scale questionnaire completed by 158 respondents from 40 major ports. The responses were aggregated using an improved Similarity Aggregation Method, and K-Means clustering was applied to classify ports into performance groups. Decision Trees were then developed to identify performance clusters and key improvement areas, while a Bayesian Network was used to explore relationships among BDAC, port operational performance, and port sustainable performance. The results indicate that ports with stronger BDAC generally achieve better operational and sustainable performance, although other contextual factors may also play important roles.
Tacit expertise (e.g. decision heuristics, risk judgement) makes organisational decision-making distinctively human, yet resides in individuals, is hard to scale, and is lost on departure. Generative AI (GenAI) offers conversational elicitation, synthesis, and adaptive scaffolding, but risks hallucination, goal drift, and helpfulness-driven agreement bias, undermining independent judgement. Building on validated problem theory, we derive four design principles grounded in socio-cognitive knowledge-work theory. They specify governance mechanisms addressing persistent transfer challenges (knowledge hiding, articulation barriers, internalisation failure) and GenAI-specific risks (knowledge distortion, tool-dependent learning, disrupted social exchange). Logical reasoning validates for alignment, applicability, coherence, completeness, and operationality. A feasibility assessment across three GenAI configurations shows most mechanisms are implementable through prompting; those requiring persistent state or platform integration are not, with configurations approximating rather than enforcing governance. We contribute prescriptive design knowledge for preserving human-centric decision-making by motivating expert sharing, enabling articulation, establishing fidelity, and ensuring applicability to successors.
Artificial intelligence (AI) promises to revolutionise radiology practices. However, the AI implementation to support clinical decision-making relies on radiologists' understanding of associated risks, as it has a pivotal role in providing outstanding healthcare outcome delivery. This study aims to explore the perception of risks related to AI and how to mitigate them from radiologists' perspective via semi-structured interviews for more in-depth information. Preliminary findings indicate that most of the literature for the most part focused on attitudes and beliefs. As a work in progress, it is significant to bridge gaps in terms of the limited of empirical studies on risk perceptions of AI and insufficient theoretical grounding. This Research-In-Progress paper has provided a research framework, methodology and anticipated contributions while also seeking insightful critique to strengthen the research argument and its significance.
We compare and contrast four types of DSS technologies using six dimensions (data, analytical environments, analytical tools, decision types, focus, and origin). Collectively, the literature indicates a clear evolutionary trajectory from MS/OR model-driven DSS to data-driven BI to Big Data Analytics to AI-enabled intelligent DSS, with XAI emerging as a necessary layer for transparency, trust, and human-computer interaction. While BI systems provide organizational visibility and analytics generate predictive insight, DSS remain essential for structuring decision problems and enabling human interaction. XAI complements these systems by addressing the interpretability challenges introduced by advanced AI techniques.
Decision support research has traditionally focused on improving choices by producing options, analyses, and recommendations. In complex and contested situations, however, the prior difficulty is often discernment: getting clear about what the situation is about, what must be understood, and what can be responsibly taken up. This paper argues that situational appropriateness cannot be judged adequately at the most immediate level of participation alone. What counts as appropriate here and now depends on how the object of concern is implicated across wider and narrower levels of participation. To address this, the paper introduces objects of discernment as what participants are answerably trying to apprehend and realise, and proposes an open-minimum analytic of seven levels of participation: individual, team/practice, project, programme, organisation, industry/profession, and society. The paper then clarifies four recurrent forms of cross-level interdependence and draws implications for answerability-centred design in decision support, including artefacts that make levels more inspectable and evaluation that distinguishes performance effects from discernment effects. The paper contributes a conceptual foundation for understanding why key situation indicator style judgements of situational appropriateness cannot be assessed only at the most immediate level of participation.
