Beyond efficiency: epistemic well-being at work under algorithmic governance

Abstract Workplace well-being research has generated extensive evidence on subjective experience and on organisational antecedents, such as leadership, climate, justice, and work design. Yet, these conditions are usually theorised as predictors of affective, evaluative, or functioning outcomes rather than as constitutive arrangements that determine whether workers can exercise professional judgement with practical consequence. In parallel, debates on artificial intelligence (AI) often frame intelligent technologies as instruments of efficiency, optimisation, or control. Focusing on systems whose outputs enter consequential organisational evaluation and decision-making (including algorithmic management, predictive analytics, automated decision-making, decision-support systems, workplace surveillance, and generative AI) this article asks whether organisational arrangements preserve or erode the capacity for judgement, contestation, and epistemic responsibility. It introduces Epistemic Well-being at Work (EWW) as an organisational condition in which individuals and collectives can exercise responsible judgement within socio-technical structures of authority. Drawing on the view of intelligent technologies as emergent epistemic regimes, this paper develops the Human Sustainability and Epistemic Well-being Framework (HSEW-F), linking AI-mediated governance, human dignity at work, and sustainable organisational well-being. The contribution is conceptual: it extends, rather than displaces, established well-being approaches by making the effective exercise of epistemic agency visible as a structural condition of human sustainability. Similar content being viewed by others 1 Introduction Algorithmic systems are not only automating tasks in contemporary organisations; they are increasingly reordering who can judge, question, and assume responsibility in everyday work. As intelligent technologies become embedded in decision-making, performance evaluation, resource allocation, and knowledge verification, the issue is no longer limited to efficiency gains or technological substitution. It also concerns how organisational authority is redistributed and whether human judgement remains meaningfully influential within AI-mediated environments (Mac Cord et al. 2026; Labraña and Ponce 2026). Consider a recruitment specialist reviewing an AI-generated candidate ranking. The system provides a score but little explanation of how variables were weighted; organisational policy treats the ranking as the default, an over-ride must be specially justified, and the human specialist remains formally accountable for the appointment. The worker may be satisfied, adequately resourced, and free from immediate strain, yet still lack a meaningful capacity to contest the knowledge claim on which the decision rests. This is the phenomenon the article seeks to make visible. Workplace well-being research has generated extensive evidence on how employees experience and evaluate work. Hedonic approaches emphasise affect and satisfaction, whereas eudaimonic approaches examine meaning, growth, autonomy, and optimal functioning (Diener et al. 2018; Ryan and Deci 2001). These traditions have produced robust measures and important interventions. The present critique does not deny those contributions or claim that well-being scholarship ignores organisations. The target is more specific. Studies already examine leadership styles, organisational climate, fairness systems, human resource practices, work design, and support as conditions associated with well-being (Diep and Horváthová 2025; Zhang et al. 2021). The limitation lies in how these conditions are usually positioned analytically: as antecedents, resources, or moderators that explain an individual affective, evaluative, or functioning outcome. Less attention is paid to organisational arrangements as constitutive of whether a worker’s reasons can enter, challenge, and alter consequential decisions. The missing level is, therefore, not “the organisation”, but the epistemic architecture of organisational authority. In this article, AI refers to intelligent technologies whose outputs enter organisational evaluation or decision-making. The scope includes algorithmic management and performance analytics, predictive scoring and automated decision-making, professional decision-support systems, workplace surveillance and people analytics, and generative AI when its outputs are used as knowledge claims or decision inputs. These technologies are not treated as equivalent: they differ in opacity, autonomy, purpose, and consequences. Their common relevance is that organisations may grant their outputs practical standing in allocating resources, assessing performance, framing problems, or authorising action. Debates on these systems have often approached them either as sources of efficiency and augmentation or as sources of discrimination, surveillance, and intensified control (Brynjolfsson and McAfee 2014; Kellogg et al. 2020; Martínez-Arvizu et al. 2025). Sergeeva et al. (2026) offer a more directly relevant lens: intelligent technologies do not merely supply information, but generate knowledge claims grounded in computational inference and may establish competing epistemic regimes within organisations. Their account clarifies how criteria of validity and authority are reconfigured. The present article builds on this regime-based view by asking a distinct question: what happens to workplace well-being when workers remain accountable for decisions but their situated judgement loses organisational standing? Research on algorithmic management has generated important insights into labour control, worker autonomy, and employment conditions (Benlian et al. 2022). Still, the intersection between workplace well-being and the effective exercise of epistemic agency remains underdeveloped. When organisational architectures subordinate judgement to opaque metrics, automated recommendations, or unexplained performance indicators, the issue extends beyond employee sentiment. It concerns whether workers can produce, evaluate, contest, and justify knowledge claims with practical consequence, who is authorised to challenge system outputs, and whether responsibility remains meaningfully human. This paper, therefore, addresses the following foundational question: How should workplace well-being be conceptualised when the core issue is not solely how employees feel, but the organisational conditions that enable or erode their capacity for professional judgement, epistemic agency, and epistemic responsibility under contemporary regimes of governance? This question acquires urgency in the current phase of organisational transformation. As algorithmic systems increasingly participate in cross-border decision-making, performance evaluation, and resource allocation, the redistribution of epistemic authority is no longer marginal but systemic. In multinational contexts characterised by institutional heterogeneity and regulatory asymmetries, such shifts may occur without explicit deliberation about their broader implications. If left unexamined, organisations risk consolidating architectures of efficiency that progressively weaken the conditions for responsible professional judgement. The erosion of professional judgement under algorithmic governance is not merely an internal organisational issue; it also bears on wider questions of accountability, legitimacy, and the human shaping of AI-mediated institutions. In response, this paper advances four theoretical contributions. First, it identifies a boundary in dominant well-being approaches: organisational conditions are extensively studied as antecedents of experience, but less often as constitutive arrangements governing the practical standing of judgement. Second, it distinguishes intelligent technologies from algorithmic governance and explains how the former become part of the latter when organisations grant their outputs authority in consequential processes. Third, it formally introduces Epistemic Well-being at Work as the organisational conditions under which workers can sustain integrity of judgement, effective epistemic agency, and recognised responsibility. Fourth, it proposes the Human Sustainability and Epistemic Well-being Framework as an initial architecture linking AI-mediated governance, human dignity at work, and sustainable organisational well-being. Rather than replacing established well-being theories, the article adds a structural analytical level. If well-being is examined as lived experience and optimal functioning, it should also be possible to ask whether organisational authority structures preserve the capacity to deliberate, contest, and act responsibly when intelligent systems participate in consequential decisions. The remainder of this paper is organised as follows. Section 2 specifies the boundary of the critique of workplace well-being research. Section 3 differentiates intelligent technologies from AI-mediated governance and identifies the systems within scope. Section 4 introduces Epistemic Well-being at Work and delineates its conceptual components. Section 5 presents the Human Sustainability and Epistemic Well-being Framework. Sections 6, 7, 8 and 9 develop theoretical propositions, integrated implications, organisational design principles, and a research agenda with explicit limitations. 2 Conceptual limits in prevailing approaches to workplace well-being Discussion of workplace well-being has advanced rapidly and has generated sophisticated accounts of both individual experience and organisational antecedents (Sonnentag et al. 2023). The concern developed here is, therefore, not that the field lacks organisational variables. It is that those variables are commonly used to explain well-being as an outcome, while the authority structures through which people judge, disagree, and assume responsibility remain analytically in the background (Salazar-Altamirano et al. 2025). This section, therefore, performs a boundary-setting function. Section 2.1 clarifies the analytical placement of organisational conditions in hedonic, eudaimonic, and happiness-oriented approaches. Section 2.2 distinguishes well-being as an outcome from well-being as a structural condition and contrasts instrumental managerial objectives with epistemic well-being. Section 2.3 explains why epistemic agency and judgement cannot be reduced to cognition, autonomy, or general agency. 2.1 Hedonic, eudaimonic and happiness-oriented approaches Workplace well-being is typically organised around two major traditions, hedonic and eudaimonic, which have generated partially distinct research agendas and validity criteria (Bravo-Sanzana et al. 2025). The hedonic tradition privileges the subjective experience of pleasure, positive affect, and satisfaction, with parsimony and empirical comparability as its principal strengths within the subjective well-being paradigm (Diener 1984). The eudaimonic tradition emphasises optimal functioning, purpose, autonomy, growth, and fulfilment, resisting the reduction of well-being to feeling good alone (Ryff 1989). A widely cited synthesis acknowledges complementarities between both traditions, yet the debate remains anchored in individual psychology (Ryan and Deci 2001). The critical issue is not that these traditions are insufficient per se, nor that they disregard leadership, climate, justice, work design, or social relations. The issue is their usual ontological and analytical placement: organisational conditions predominantly enter models as causes or resources, whereas the reported or functioning individual remains the location in which well-being is ultimately observed. Three assumptions become problematic when this model is carried into AI-mediated decision environments. First assumption: well-being is operationally located primarily in individual experience. Even when eudaimonic dimensions are incorporated, well-being is generally observed through the worker’s affect, evaluation, or functioning. Models of happiness at work commonly integrate satisfaction, affect, and positive attitudes as core components (Fisher 2010). Organisational context is not absent, but its role is usually to influence these outcomes rather than to constitute whether professional judgement, reasoned disagreement, or responsibility carries practical authority. Second assumption: epistemic agency is rarely treated as a governance relation. Hedonic approaches include cognitive evaluations, and eudaimonic frameworks incorporate autonomy or environmental mastery. These are important, but they do not usually examine how professional reasoning is embedded in standards of evidence, epistemic hierarchies, performance metrics, and institutional pressures (Delle Fave et al. 2010). When automated recommendations or algorithmic metrics are granted presumptive authority, the change is not merely cognitive or operational. It alters what can be defended as valid judgement and the organisational costs attached to doing so. Third assumption: well-being becomes a manageable product without ontological clarification. Approaches such as happiness at work and happiness management advocate deliberate organisational practices to foster well-being, engagement, and sustainable outcomes (Ravina-Ripoll et al. 2024). This shift beyond the isolated individual is valuable. A critical distinction is nevertheless required between the language of well-being and the objectives of managerial programmes. In practice, organisations may seek legitimate but narrower outcomes—lower sickness absence, reduced burnout, improved retention, or continuity of performance—without committing to well-being in a deeper sense of autonomy, agency, or responsibility. The theoretical risk is to treat these outcomes as evidence that the organisational conditions of responsible judgement have also been secured. To render this fracture visible without oversimplification, Table 1 summarises the theoretical strengths and structural blind spots of dominant approaches when applied to contemporary organisational settings. The purpose of this critique is not to dismiss these approaches. Their cumulative contribution is substantial, and their attention to organisational antecedents is explicit. The narrower claim is that, in digitally mediated settings, scales and intervention models can describe experience while overlooking whether professional reasons have institutional standing. EWW addresses this omitted structural level rather than adding another affective or cognitive variable. At this point, progress cannot be achieved merely by adding another psychological variable to well-being. What is required is a distinction between well-being as a psychological outcome and well-being as a structural condition. 2.2 Well-being as outcome versus well-being as structural condition If the previous section showed that workplace well-being is predominantly observed through affective, evaluative, or functioning outcomes, the analysis must now move to a different plane: is well-being only an outcome that emerges from organisational conditions, or may it also name a condition that enables responsible professional action? This is an ontological distinction, because it determines which organisational phenomena become visible. In mainstream organisational research, well-being typically appears as a dependent variable or, at best, as a mediator between organisational practices and performance outcomes (Galván-Vela et al. 2026). Within the Job Demands–Resources model, for example, well-being is understood as the result of the balance between demands and resources, explaining engagement, burnout, and performance (Bakker and Demerouti 2007). This framework has proven empirically productive. However, it presupposes that well-being is a state attained after interaction with work conditions rather than a human infrastructure sustaining the possibility of acting with professional judgement. Similarly, flourishing-based approaches continue to treat well-being as an achievement or optimal functioning state (Huppert and So 2011). Although conceptually richer than purely hedonic accounts, they share an implicit structure: well-being is something individuals possess to varying degrees. What remains insufficiently theorised is the possibility that, in complex organisational contexts, well-being may also operate as a prior structural condition enabling or constraining judgement, responsible deliberation, and reasoned resistance. This distinction becomes clearer when two analytical planes are differentiated, as summarised in Table 2. This distinction does not invalidate existing models; rather, it clarifies their analytical level. The difficulty arises when findings about outcome well-being are translated into structural prescriptions, as if the two levels were interchangeable. Increasing resources or positive emotions may improve health, satisfaction, or engagement, but it does not necessarily preserve the organisational standing of professional judgement (Fredrickson 2001). An organisation may report high satisfaction and operate credible programmes to reduce stress or burnout while still making automated decisions difficult to question. Under an outcome-oriented framework, such a context may reasonably be classified as supportive. Under a structural framework, it may simultaneously reveal erosion of the epistemic infrastructure of work. The two assessments are not mutually exclusive, because they address different objects. This also clarifies the managerial issue. Preventing illness, exhaustion, or absence is an important organisational responsibility, but it is not equivalent to protecting epistemic well-being. A programme may reduce strain while leaving intact a decision architecture in which system outputs are presumptively correct and human over-rides are penalised. The critical question is not whether managerial motives are sincere, but whether the governed object is employee functioning alone or also the institutional capacity for accountable judgement. Treating well-being solely as an outcome also risks three inferential errors: - 1. Functional equivalence error: assuming positive emotions ensure responsible agency (An 2025). - 2. Linear causality error: presuming that increasing resources automatically enhances deliberative capacity (John et al. 2023). - 3. Structural neutrality error: overlooking how decision architectures may redistribute epistemic authority independently of emotional climate (Velotto 2025). These are not methodological flaws, but limits of inference across analytical levels. If well-being can also be examined as a condition enabling judgement (Abbas et al. 2022), the question expands from which practices improve positive experience to which institutional arrangements preserve or erode the capacity to reason and respond under responsibility (Guenduez et al. 2024). This shift leads directly to the next step in the argument. Certain contemporary organisational developments do not merely function as tools or resources; they operate as regimes of governance that reconfigure authority, responsibility, and criteria of validity. 2.3 The invisibilisation of epistemic agency and judgement The core gap is not that organisational research neglects cognition. Decision-making, learning, biases, information processing, autonomy, and knowledge have all been studied extensively (Yusif and Hafeez-Baig 2024). The problem is that cognition is commonly analysed as an individual mental process, while knowledge is often analysed as a strategic resource. Less attention is given to epistemic agency as a structural relation embedded in organisational regimes of knowledge production, validation, and authority (Gernigon et al. 2023). Classical organisational theory recognised that organisations exist, precisely because individual rationality is bounded (Simon 1947). However, subsequent developments often operationalised this insight in terms of efficiency, routines, and heuristics rather than the integrity of professional judgement within authority structures. Even approaches grounded in the knowledge-based view emphasise knowledge creation and transfer as strategic resources (Grant 1996), while rarely interrogating what occurs when criteria of validity or epistemic authority are structurally transformed. Research on algorithmic management shows that digital systems can intensify control, redistribute autonomy, and reconfigure supervision (Kellogg et al. 2020). These analyses are essential, but autonomy and control do not fully capture whether a professional’s reasons can challenge the knowledge claim embedded in an automated recommendation or metric. The invisibilisation thus operates at two complementary levels. First, workplace well-being research centres on emotions, attitudes, and evaluations, while organisational knowledge research prioritises efficiency and competitive advantage. Between these domains lies a gap: limited attention has been given to how organisational conditions shape the possibility of exercising responsible judgement under structured authority. Second, when cognition appears in well-being models, it typically does so as perception, for example, perceived justice or perceived support, rather than as normative judgement. The distinction is substantial. Perceiving fairness is not equivalent to being able to challenge a decision. Feeling supported is not equivalent to dissenting without sanction. Reporting satisfaction is not equivalent to assuming epistemic responsibility for outcomes. To clarify the distinction, Table 3 summarises the principal levels at which cognition and knowledge are treated in organisational research and adds the epistemic level proposed here. The proposed fourth level is epistemic rather than merely cognitive. It concerns not only the capacity to think, but the organisational standing of knowledge claims and reasons. This terminology is more precise than autonomy or agency alone. Autonomy denotes discretion over action; general agency denotes the capacity to act and effect change. Epistemic agency concerns the ability to contribute to, evaluate, contest, and justify knowledge claims—and for those reasons to carry practical weight in a decision. A worker may retain discretion over task execution yet be required to accept an opaque risk score, or may be invited to speak, while their reasons have no decisional effect. EWW, therefore, incorporates autonomy and agency but is not reducible to either. This absence becomes particularly salient in digitally mediated decision environments. Contemporary systems do not merely process information more rapidly; they also reshape who defines what counts as evidence, which criteria prevail in decision-making, and who bears responsibility for consequences (Peeters 2020). When such arrangements are naturalised as technological support, the redistribution of epistemic authority escapes the analytical frame of workplace well-being. Within this ecosystem, erosion of judgement does not necessarily manifest as immediate stress or dissatisfaction. It may appear as growing dependence, declining professional disagreement, excessive standardisation of criteria, or uncritical internalisation of metrics (Lorenz 2025). These phenomena are rarely interpreted as indicators of well-being. They are often framed as efficiency, alignment, or technological adaptation. From a structural perspective, however, they may signal weakening of the epistemic infrastructure of human work. The problem, therefore, cannot be formulated solely in terms of positive affect, demand–resource balance, autonomy, or control. Formal discretion may remain, while professional judgement loses effective weight. What is at stake is the preservation of responsible epistemic agency within organisational structures that define practical truth and decisional legitimacy. This recognition completes the conceptual boundary developed in this section. The next step is to distinguish intelligent technologies from the governance arrangements through which organisations authorise their outputs. 3 From intelligent technologies to AI-mediated governance The argument now requires a distinction between intelligent technologies and governance. AI systems are technical and socio-technical artefacts. Algorithmic governance refers to the institutional arrangements through which organisations give system outputs a role in structuring behaviour, allocating resources, evaluating performance, or authorising decisions. AI is, therefore, not synonymous with governance; it becomes part of governance through organisational design, policy, incentives, and practice. Instrumental accounts frame intelligent technologies as enhancing efficiency, optimising decisions, automating tasks, or augmenting human capabilities (Parycek et al. 2023). Critical accounts emphasise productivity effects, surveillance, employment relations, or intensified control (Kellogg et al. 2020). These perspectives remain valuable, but they do not always capture how system outputs acquire standing as claims about what is likely, valid, optimal, or true. Sergeeva et al. (2026) address this problem by conceptualising intelligent technologies as sources of knowledge claims grounded in computational modelling, statistical inference, and data infrastructures. Such claims can compete with professional expertise and contribute to new epistemic regimes governing how knowledge is produced, evaluated, and authorised. This perspective directly supports the paper’s shift beyond a tool view. EWW extends it by examining the human well-being implications of participating in those regimes. The distinction is consequential. A system may be technically identical across two organisations but governed differently. Where its output is advisory, contestable, and linked to identifiable human responsibility, it may support judgement. Where it becomes an unquestioned default reinforced by incentives and penalties, it may displace judgement. The governance arrangement—not the label “AI” alone—determines the relevant epistemic conditions. Scope of the intelligent technologies considered Table 4 distinguishes the principal forms considered in this article. The categories overlap in practice, and the table does not imply that their effects are uniform. It identifies the mechanism through which each can acquire organisational authority and provides an illustrative EWW concern. 3.1 Reconfiguration of responsibility In AI-mediated decision contexts, responsibility may become diffused (Rangel-Lyne and Salazar-Altamirano 2025). When automated recommendations guide action, professionals may frame outcomes as originating from the system, even though formal accountability remains human. The problem is amplified when organisations require compliance with outputs but provide weak mechanisms for explanation or challenge. Research on automation documents phenomena, such as automation bias, whereby individuals over-rely on technological outputs (Parasuraman and Riley 1997). However, the organisational issue extends beyond cognitive bias. It concerns structural incentives: what rewards or sanctions accompany the questioning of automated outputs? What metrics legitimise dissent? What organisational signals define compliance as prudence? Where organisational architectures reward alignment with automated outputs and penalise deviation, the capacity for independent judgement may erode progressively (Naser 2025). Emotional well-being may remain stable, while the structural conditions for responsible judgement weaken. 3.2 Redistribution of agency The integration of intelligent systems may also redistribute epistemic authority within organisations (Herrmann and Pfeiffer 2022). Part of the legitimacy previously attached to professional or hierarchical judgement may shift towards data-driven models whose criteria are not transparent to users (Liang et al. 2022). The organisation does not lose control; rather, it reconfigures, where valid knowledge is presumed to reside. Kellogg et al. (2020) show that algorithmic management can intensify control and reduce professional discretion. However, beyond control lies a more fundamental issue: the redefinition of what counts as valid evidence and who is authorised to challenge it. Algorithmic regimes do not merely coordinate tasks; they structure the architecture of legitimate knowledge. This shift has direct implications for epistemic agency. Professional judgement may persist formally but lose effective influence when automated outputs are treated as the default starting and ending point. Agency is not eliminated; it is recalibrated within boundaries established by data infrastructures, optimisation logics, and performance metrics. 3.3 Reconfiguration of decisional authority Finally, authority becomes not only interpersonal or hierarchical but systemic. Metrics, dashboards, and automated recommendations may acquire a status of technical objectivity that renders them difficult to contest. Authority becomes less visible, yet no less consequential. As Danaher et al. (2017) argue, algorithmic governance may operate through opaque mechanisms that shape decisions without explicit human intervention. Within organisational contexts, this means that professional deliberation can become subordinated to outputs whose internal logic remains partially inaccessible (Frémeaux and Voegtlin 2022). When such subordination is normalised in the name of efficiency, its implications for well-being are rarely examined. The issue is not whether algorithms improve performance, but whether organisational structures preserve the conditions for responsible judgement under redistributed authority. Conceptualising intelligent technologies within AI-mediated governance, therefore, makes visible how responsibility, epistemic agency, and legitimacy are reorganised in contemporary workplaces. This position is neither technophilic nor technophobic, and it does not assume that every intelligent technology produces the same organisational effect. It directs attention to the arrangements through which computational claims are authorised, contested, and connected to human responsibility. At this juncture, the conceptual gap identified in the previous sections becomes explicit. If workplace well-being is reduced to psychological states and intelligent systems structurally reorganise epistemic authority, a construct is required to capture their intersection. The following section introduces that construct: epistemic well-being at work. 4 Epistemic well-being at work (EWW) The preceding sections identified a specific boundary in well-being research and distinguished intelligent technologies from the governance arrangements that authorise their outputs. This section introduces a construct that captures their intersection: the organisational conditions under which professionals can exercise responsible judgement within structured regimes of epistemic authority. The section unfolds in four steps: formal definition, conceptual differentiation, structural components and consequences of erosion. 4.1 Formal definition EWW is defined as the organisational condition under which individuals and collectives can exercise responsible professional judgement, effective epistemic agency, and recognised epistemic responsibility within structures of authority that legitimate reasoned deliberation and contestation. This definition contains five non-substitutable elements: - 1. Organisational condition, not an individual psychological state. - 2. Exercise of professional judgement, rather than subjective evaluation. - 3. Effective epistemic agency, understood as the practical capacity to produce, evaluate, contest, and act upon knowledge claims with consequence. - 4. Epistemic responsibility, meaning the ability to answer for the validity of a decision or criterion. - 5. Structures of authority and governance that enable or constrain these capacities. Unlike hedonic or eudaimonic well-being, EWW does not describe how individuals feel or how satisfied they are. It addresses whether organisational architecture preserves the conditions for responsible professional judgement under structured authority. 4.2 Conceptual differentiation from related constructs The legitimacy of a foundational construct depends on clear boundaries. EWW overlaps with well-being, autonomy, agency, justice, and voice, but its distinctive object is whether professional reasons and knowledge claims have practical standing within organisational authority. Table 5 summarises these differences. EWW may coexist with high satisfaction or engagement levels, yet it is not reducible to them. An organisation may display committed and emotionally positive employees while structurally constraining professional judgement. Psychological safety, autonomy, and general agency remain necessary adjacent concepts. EWW adds the requirement that professional judgement not only be expressible or discretionary, but institutionally consequential. It concerns whether reasons can enter the decision process, challenge computational claims, and support accountable action. 4.3 Conceptual components At this stage, EWW is a conceptual construct rather than a validated scale. It rests on three interdependent components that require future empirical examination. Integrity of judgement: The capacity to formulate, sustain and justify professional decisions on reasoned grounds, even under structural pressure. Integrity implies that dissent does not incur disproportionate organisational cost and that deliberation remains institutionally legitimate. Effective epistemic agency: Thinking alone is insufficient. Organisational architecture must not systematically neutralise human judgement in favour of automated outputs. Agency is effective when professional reasoning can interrogate knowledge claims and meaningfully influence the final decision (Roumbanis 2025). Recognised epistemic responsibility: Actors must be able to answer for their decisions in terms of reasons, not solely in terms of metric compliance. This component connects with broader debates on responsibility in automated systems (Danaher et al. 2017). These components are not psychological traits. They are structural properties expressed through governance practices, process design, incentive systems and criteria of legitimacy. 4.4 When EWW erodes The weakening of Epistemic Well-being at Work does not necessarily manifest as immediate dissatisfaction or emotional strain. Erosion may be gradual and structurally embedded. Indicative manifestations include: - Increasing epistemic dependence on automated recommendations. - Reduced professional disagreement due to structural costs of contestation. - Uncritical internalisation of metrics, where compliance replaces deliberation. - Displacement of responsibility, with decisions attributed to systems rather than accountable actors. Such dynamics may coexist with positive satisfaction indicators. However, from a human sustainability perspective, they signal a weakening of the epistemic infrastructure that sustains professional dignity and responsible judgement. EWW is neither a moral category nor a technological critique. It is an analytical construct that allows scholars to identify when organisational architectures preserve or undermine the structural conditions for accountable professional judgement. Having defined the construct, the next step is to situate EWW within a broader architecture linking AI-mediated governance, human dignity, and sustainable organisational well-being. Measurement and empirical testing remain future tasks. The following section introduces the architecture: the Human Sustainability and Epistemic Well-being Framework, in which EWW serves as the integrative node of a coherent conceptual model. 5 Human sustainability and epistemic well-being framework (HSEW-F) Up to this point, the article has specified the target of its well-being critique, differentiated intelligent technologies from their governance arrangements, and introduced EWW. The task now is integrative. The HSEW-F situates EWW within a provisional architecture connecting AI-mediated governance, human dignity, and sustainable organisational well-being under a human-sustainability logic. 5.1 From efficiency to human sustainability In contemporary management debates, sustainability has evolved beyond environmental and financial metrics to incorporate human capital and social legitimacy (Spanuth and Urbano 2023). However, even advanced organisational sustainability frameworks tend to focus on performance indicators, engagement levels, or aggregated well-being outcomes, without interrogating the structural conditions under which individuals exercise responsible agency within technologically mediated systems. From this perspective, organisational sustainability becomes fragile when the epistemic capacity of those operating within it is progressively eroded. Efficiency and positive affect are insufficient if the organisational architecture undermines the integrity of professional judgement. Long-term legitimacy depends not only on output optimisation, but also on preserving the conditions for accountable reasoning under structured authority. 5.2 Conceptual architecture of the framework As an initial integrative heuristic, the HSEW-F is organised around four inter-related nodes: - 1. Organisational algorithmic governance - 2. Human dignity at work - 3. Epistemic well-being at work (EWW) - 4. Sustainable organisational well-being These elements do not form a validated causal chain. They constitute a relational architecture in which EWW links governance arrangements to the recognition of workers as deliberative and accountable agents (see Fig. 1). The fourth node is deliberately framed as sustainable organisational well-being rather than happiness. Happiness is a valuable affective or evaluative outcome, but the framework’s endpoint is broader: durable organisational conditions supporting health, dignity, agency, and responsible participation. 5.3 Algorithmic governance as a structural condition AI-mediated governance is understood here as the set of institutional arrangements through which intelligent systems participate in decision-making, resource allocation, performance evaluation, or knowledge production. Technology is not itself the governance regime; organisational policies, incentives, appeal mechanisms, and responsibility structures determine how its outputs acquire authority. Depending on design and governance choices, algorithmic governance may: - support deliberative processes - marginalise professional disagreement - privilege metric compliance - obscure decision criteria This may be observed, for example, in recruitment screening systems, performance evaluation dashboards, clinical decision-support tools, or university assessment systems, where system outputs increasingly shape what counts as a valid or contestable decision. Algorithmic governance thus constitutes the structural environment within which EWW may be preserved or weakened. 5.4 EWW as a structural intermediary EWW operates as an intermediary structural condition between governance architecture and sustainable organisational outcomes. It does not denote emotional experience, but the preservation of three capacities: integrity of judgement, effective epistemic agency, and recognised epistemic responsibility. When these capacities are maintained, organisations do more than protect professional autonomy. They sustain human dignity at work, understood as the recognition of individuals as reasoning agents capable of deliberation and accountability. In this sense, EWW bridges organisational theory and applied ethics. Dignity is not maintained solely through respectful interaction; it also requires structural arrangements that legitimate reasoned deliberation within decision systems. 5.5 Human dignity and sustainable organisational well-being Organisational dignity research emphasises that work is a central domain of moral recognition (Pirson 2017). In technologically mediated environments, dignity may be compromised not only through overt mistreatment, but also through the progressive subordination of human judgement to opaque systems. Sustainable organisational well-being is broader than episodic happiness or short-term satisfaction. It refers here to the durability of conditions supporting health, dignity, agency, meaningful participation, and responsibility. Favourable affective indicators may coexist with epistemic fragility; conversely, preserving EWW is not sufficient by itself to secure every dimension of well-being. The HSEW-F, therefore, proposes a relational logic: - a. algorithmic governance configures the decision environment; - b. EWW determines whether responsible judgement remains structurally viable; - c. human dignity depends on that viability; - d. sustainable organisational well-being becomes more plausible when these conditions align. 5.6 Integrative logic of the framework At the architectural level, the framework advances five connected claims. Contemporary organisations increasingly operate through AI-mediated decision architectures; these architectures can redistribute epistemic authority; that redistribution may strengthen or weaken EWW; preserving EWW supports recognition of workers as deliberative and accountable agents; and such recognition is one structural contribution to sustainable organisational well-being. The framework does not claim that technology is inherently detrimental, that human judgement is invariably superior, or that EWW is sufficient for overall well-being. Its narrower claim is that human sustainability requires alignment between accountability and the practical capacity to reason within complex authority structures. With the framework established, the next step is logical ordering rather than empirical testing. The following section formulates a set of theoretical propositions derived from this architecture, intended to guide future research without converting them into statistical hypotheses. 6 Theoretical propositions The purpose of this section is not to anticipate empirical testing nor to formulate statistically testable hypotheses. Its function is more foundational: to logically order the implications derived from the HSEW-F. The propositions do not predict coefficients or measurable magnitudes. Instead, they articulate necessary theoretical relations if the framework is internally coherent. Each proposition is formulated conditionally. The aim is not to assert deterministic causality, but to clarify structural plausibility under specific organisational configurations. P1. Governance and epistemic conditions P1: If organisational governance reallocates epistemic authority towards automated systems without explicit mechanisms for human deliberation, the integrity of professional judgement is likely to weaken as a structural condition. This proposition follows from the conceptual shift outlined in Sect. 3. When decision architectures privilege automated outputs as primary sources of legitimacy, epistemic agency may become secondary. The proposition does not assume that intelligent systems inevitably undermine judgement; organisational design and the degree of contestability are decisive. P2. Epistemic agency and human dignity P2: To the extent that EWW is preserved, human dignity at work is more likely to be sustained as the effective recognition of individuals as deliberative agents. Organisational dignity extends beyond respectful treatment or formal equity. It depends on the structural possibility of exercising accountable judgement. EWW, therefore, operates as a precondition for dignity to function as more than symbolic recognition. P3. Silent erosion and affective well-being P3: High levels of affective well-being may coexist with low levels of EWW when organisational architectures restrict epistemic agency without generating immediate emotional discomfort. This proposition guards against empirical conflation. EWW does not collapse into satisfaction or engagement. Its erosion may be gradual and structurally embedded, remaining undetected by affect-based measures in the short term. In settings, such as university assessment systems or clinical decision-support environments, employees may report acceptable levels of satisfaction while nevertheless experiencing diminished scope to question system-generated outputs. P4. Institutionalised deliberation and structural preservation P4: If organisations institutionalise formal mechanisms for reasoned deliberation regarding automated decisions, EWW is likely to strengthen as a structural condition. The emphasis lies not on individual attitudes but on institutional design. Structurally protected spaces for questioning, revising, and justifying algorithmically mediated decisions function as preservation mechanisms for epistemic integrity. P5. EWW and sustainable organisational well-being P5: Sustainable organisational well-being is more plausible when governance arrangements preserve EWW than when favourable outcomes rest solely on affective states or instrumental incentives. This proposition does not treat EWW as a sufficient cause of well-being. It states that well-being claims are structurally more robust when workers retain meaningful capacity to judge, contest, and answer for consequential decisions. P6. Distributed responsibility and organisational coherence P6: When decisional responsibility becomes diffusely distributed between automated systems and human actors without normative clarity, EWW is likely to fragment, thereby weakening organisational coherence. Diffuse accountability is not merely an ethical concern; it is also an epistemic one. Where responsibility lacks clear attribution, the integrity of professional judgement becomes structurally unstable. This may become especially visible in employee-facing contexts, such as risk scoring or automated recommendation systems, where action is guided by outputs, while accountability remains opaque. Collectively, these propositions delineate the conceptual scope of EWW, connect AI-mediated governance with dignity and sustainable organisational well-being, and provide a conditional foundation for empirical research. They are theoretical propositions rather than validated causal claims. More importantly, the propositions demonstrate that EWW is not an abstract normative category but a structural component necessary for understanding human sustainability in contemporary organisations. With the framework and its derived propositions established, the next step is evaluative rather than additive. The following section examines how incorporating EWW reshapes existing organisational theories and exposes their limitations when the epistemic dimension remains untheorised. 7 Theoretical implications: three connected shifts Rather than treating workplace well-being, organisational behaviour, and AI governance as separate domains, EWW connects them through a single question: whether decision architectures preserve the practical standing of human judgement. The implications can, therefore, be organised around three connected shifts. 7.1 From organisational antecedents to constitutive decision conditions Established well-being scholarship remains indispensable for explaining affect, satisfaction, functioning, and strain. EWW does not dispute the role of leadership, climate, justice, resources, or work design; it changes their analytical placement when decisions are AI-mediated. Organisational conditions are not only antecedents of individual outcomes. They can be constitutive of whether workers are able to give reasons, contest evidence, and influence consequential decisions. Favourable self-reports cannot, by themselves, establish that a work system is humanly sustainable. A recruitment unit, for example, may reduce workload and burnout while still requiring staff to accept an opaque ranking model. The intervention may improve outcome well-being without preserving the structural conditions of judgement. EWW identifies this non-equivalence rather than treating experiential and structural well-being as competitors. 7.2 From AI tools to contested epistemic regimes Sergeeva et al. (2026) conceptualise intelligent technologies as sources of knowledge claims that can form competing epistemic regimes. EWW adds a well-being question to that perspective. When computational criteria gain authority, the issue is not only accuracy, bias, or transparency, but how professionals are positioned in relation to those claims: as accountable judges, procedural approvers, or passive recipients. AI-mediated governance is, therefore, a relational organisational achievement, not a property of technology alone. The same predictive model may support EWW, where its criteria are intelligible, contestable, and connected to human authority, but weaken EWW, where output compliance is rewarded and over-rides are penalised. The framework thus connects system design with incentives, legitimacy, and responsibility. 7.3 From formal autonomy and voice to effective epistemic agency Autonomy, general agency, voice, and psychological safety are necessary adjacent concepts, but they do not resolve the specific problem. Autonomy concerns discretion; general agency concerns capacity to act; voice concerns expression. Epistemic agency concerns the ability to produce, evaluate, contest, and justify knowledge claims and for those reasons to carry practical weight. An employee may be free to speak yet unable to alter a decision, or may choose how to perform a task while being required to accept the system’s classification. EWW, therefore, implies that well-being analysis should include the structural efficacy of judgement, organisational behaviour should distinguish motivated compliance from deliberative agency, and governance quality should be assessed partly by whether responsibility remains answerable in human reasons. It does not replace existing theories; it supplies an integrative level for settings in which intelligent technologies reshape the grounds of organisational knowing. 8 Organisational design implications: preserving epistemic agency If EWW is a structural condition, its practical implications operate at the level of decision architecture rather than as another wellness intervention. The objective is not to prescribe a universal checklist, but to identify three connected domains through which organisations can align technological use with accountable human judgement. 8.1 Contestability and deliberative authority Consequential AI-mediated recommendations should remain open to reasoned review through explicit over-ride, appeal, and escalation routes. A formal right to speak is insufficient unless an alternative judgement can affect the outcome and does not trigger disproportionate penalties. In recruitment, for example, a specialist should be able to challenge an automated ranking, document context omitted by the model, and obtain review by an accountable decision-maker. Metrics should, therefore, function as evidence rather than substitutes for judgement. Performance systems should record justified deviations and evaluate the quality of reasoning, not treat deviation itself as error. This design principle preserves the possibility that professional expertise can correct rather than merely execute a computational claim. 8.2 Traceability and accountable responsibility Traceability should be calibrated to role and consequence. Workers need sufficient information about inputs, assumptions, uncertainty, and limits to understand what claim is being made and where human intervention is expected. This differs from demanding complete technical transparency in every case; it requires actionable intelligibility for responsible participation. Responsibility should also remain identifiable and aligned with authority. Where organisations require humans to sign off AI-assisted decisions, they must provide corresponding capacity to interrogate and revise those outputs. Accountability without epistemic authority creates a structurally incoherent arrangement in which responsibility is retained, while judgement is displaced. 8.3 Alignment with human sustainability Well-being and sustainability commitments should be examined against actual decision architecture. Reduced burnout, a positive climate, or high engagement do not compensate for systematic exclusion of professional reasons from consequential decisions. Conversely, EWW does not require rejecting automation; it requires governance that preserves the human capacity to judge, contest, and answer for outcomes. Implementation can be examined through a connected set of questions: Which outputs receive presumptive authority? Who may challenge them? What evidence is required? What organisational costs attach to over-ride? Who remains accountable? Together, these questions translate EWW into a design lens while preserving sensitivity to sector, risk, and technological form. 9 Research agenda and theoretical limitations The HSEW-F is conceptual and requires empirical and theoretical development. A coherent programme should distinguish construct development, technology-specific mechanisms, and boundary conditions rather than assume that all AI-mediated work produces the same effects. 9.1 Construct development and measurement Operationalisation should begin with qualitative and comparative research identifying how integrity of judgement, effective epistemic agency, and recognised responsibility are enacted across occupations. Scale development may then assess employees’ perceptions of structural conditions, but perception measures should be triangulated with governance artefacts and practices, including over-ride rules, audit trails, incentive systems, appeal routes, and actual decision influence. Confirmatory factor analysis, discriminant validity, measurement invariance, and longitudinal designs will be necessary to establish whether EWW is distinct from autonomy, psychological safety, justice, and established well-being constructs. 9.2 Technology-specific and comparative mechanisms The broad scope of intelligent technologies is a starting boundary, not a claim of equivalence. Future research should compare algorithmic management, automated decision-making, professional decision-support, surveillance analytics, and generative AI because each may redistribute authority through different mechanisms. Comparative case studies, ethnographies, experiments, and process research can examine how computational claims gain legitimacy, how workers accommodate or resist them, and how responsibility is negotiated in hybrid arrangements. Cross-national and institutional comparison is also essential. Regulatory capacity, labour protections, professional authority, technological dependence, and cultural expectations may shape whether contestation is feasible. Research in emerging economies and Global South contexts can prevent the framework from assuming highly regulated or professionally privileged settings as the default. 9.3 Boundary conditions and limitations Several limitations bound the present contribution. First, the article is conceptual and provides no empirical evidence that EWW predicts dignity, legitimacy, or sustainable organisational well-being; the propositions remain theoretically plausible rather than validated. Second, EWW is formulated at the organisational level but may be experienced unevenly across hierarchy, occupation, employment status, and access to technical expertise, making aggregation a substantive empirical problem. Third, the focus on epistemic conditions does not imply that EWW is sufficient for overall well-being: health, income, security, workload, relationships, and material conditions remain indispensable. Fourth, this paper does not provide a legal analysis or claim that all forms of AI create uniform effects. Sectoral risk, regulation, system autonomy, data quality, and the reversibility of decisions are likely boundary conditions. Fifth, the emphasis on human judgement should not romanticise professional expertise; human decisions can be biased, inconsistent, or exclusionary, and intelligent systems may improve accuracy and reveal errors. The relevant issue is, therefore, not human versus machine superiority, but whether their interaction is governed through contestable knowledge claims and aligned responsibility. These limitations indicate, where EWW is most likely to be relevant: contexts in which intelligent outputs have consequential organisational standing and human actors remain expected to interpret, justify, or answer for outcomes. It may be less central in low-consequence automation with no meaningful judgement component. Future research should test these boundaries and examine whether EWW explains learning, error detection, ethical voice, organisational legitimacy, and durable well-being beyond adjacent constructs. 10 Conclusion The objective of this conceptual article was to articulate a structural dimension of workplace well-being that becomes salient when intelligent technologies acquire practical authority in organisational decisions. Established research already demonstrates that leadership, climate, justice, resources, and work-design shape well-being outcomes. The narrower claim advanced here is that these literatures have not yet fully theorised whether decision architectures preserve the effective exercise of professional judgement, epistemic agency, and responsibility. To address this omission, the article distinguished well-being as an outcome from well-being as a constitutive condition; specified the forms of intelligent technology within scope; and drew on the epistemic-regime perspective to explain how computational knowledge claims may reconfigure authority. It then introduced EWW and the HSEW-F, linking AI-mediated governance, human dignity, and sustainable organisational well-being. The framework does not imply that all AI systems or all organisations produce the same effects. It does not establish empirically that EWW causes dignity or well-being, and it does not treat human judgement as inherently superior. Its claim is conditional: where organisational systems expect humans to remain accountable for consequential decisions, human reasons must retain meaningful capacity to interrogate and alter the knowledge claims on which those decisions depend. EWW, therefore, complements rather than replaces affective, eudaimonic, autonomy, agency, justice, or psychological-safety constructs. Its distinctive object is the organisational standing of judgement. A workplace may be supportive and emotionally positive yet epistemically fragile; it may also use AI extensively while preserving EWW through contestability, traceability, and aligned responsibility. Future empirical research must determine whether EWW offers explanatory and predictive value beyond adjacent constructs and under which technological, occupational, and institutional conditions. The conceptual contribution is to make that question available for systematic investigation. Accordingly, the evaluation of AI-mediated work should ask not only how efficiently decisions are made, but whether those expected to answer for them retain the authority and conditions to judge. Data availability No datasets were generated or analysed during the current study. References Abbas A, Ekowati D, Suhariadi F, Hamid SaR (2022) Negative vs. positive psychology: a review of science of well-being. Integr Psychol Behav Sci 58(4):1091–1122. https://doi.org/10.1007/s12124-022-09708-1 An D (2025) Artificial reactive attitudes. AI Soc. https://doi.org/10.1007/s00146-025-02622-1 Bakker AB, Demerouti E (2007) The job demands-resources model: state of the art. J Manag Psychol 22(3):309–328. https://doi.org/10.1108/02683940710733115 Benlian A, Wiener M, Cram WA, Krasnova H, Maedche A, Möhlmann M, Recker J, Remus U (2022) Algorithmic management. Bus Inf Syst Eng 64(6):825–839. https://doi.org/10.1007/s12599-022-00764-w Bravo-Sanzana M, Casas F, Rodríguez-Rivas ME, Oriol X, Varela JJ, Miranda R, Terán-Mendoza O (2025) Instruments for measuring hedonic and eudaimonic well-being of adolescents in the Latin American school contexts: a systematic review. Child Indic Res 18(3):955–1000. https://doi.org/10.1007/s12187-025-10231-z Brynjolfsson E, McAfee A (2014) The second machine age: work, progress, and prosperity in a time of brilliant technologies. W. W. Norton & Company Danaher J, Hogan MJ, Noone C, Kennedy R, Behan A, De Paor A, Felzmann H, Haklay M, Khoo S, Morison J, Murphy MH, O’Brolchain N, Schafer B, Shankar K (2017) Algorithmic governance: developing a research agenda through the power of collective intelligence. Big Data Soc 4(2):205395171772655. https://doi.org/10.1177/2053951717726554 Delle Fave AD, Massimini F, Bassi M (2010) Hedonism and eudaimonism in positive psychology. In: Cross-cultural advancements in positive psychology, pp 3–18. https://doi.org/10.1007/978-90-481-9876-4_1 Diener E (1984) Subjective well-being. Psychol Bull 95(3):542–575. https://doi.org/10.1037/0033-2909.95.3.542 Diener E, Oishi S, Tay L (2018) Advances in subjective well-being research. Nat Hum Behav 2(4):253–260. https://doi.org/10.1038/s41562-018-0307-6 Diep TPT, Horváthová P (2025) Human resource management, employee well-being, and performance: a systematic review. Hum Syst Manag 45(1):89–108. https://doi.org/10.1177/01672533251355886 Fisher CD (2010) Happiness at work. Int J Manag Rev 12(4):384–412. https://doi.org/10.1111/j.1468-2370.2009.00270.x Fredrickson BL (2001) The role of positive emotions in positive psychology: the broaden-and-build theory of positive emotions. Am Psychol 56(3):218–226. https://doi.org/10.1037/0003-066x.56.3.218 Frémeaux S, Voegtlin C (2022) Strengthening deliberation in business: learning from Aristotle’s ethics of deliberation. Bus Soc 62(4):824–859. https://doi.org/10.1177/00076503221113816 Galván-Vela E, Martínez-Arvizu OJ, Salazar-Altamirano MA, Ravina-Ripoll R (2026) Effects of happiness management and organisational justice on organisational commitment. J Manag Dev 45:1–14. https://doi.org/10.1108/jmd-05-2025-0292 Gernigon C, Hartigh RJRD, Vallacher RR, Van Geert PLC (2023) How the complexity of psychological processes reframes the issue of reproducibility in psychological science. Perspect Psychol Sci 19(6):952–977. https://doi.org/10.1177/17456916231187324 Grant RM (1996) Toward a knowledge-based theory of the firm. Strateg Manag J 17(S2):109–122. https://doi.org/10.1002/smj.4250171110 Guenduez AA, Mergel I, Schedler K, Fuchs S, Douillet C (2024) Institutional work in smart cities: interviews with smart city managers. Urban Gov 4(1):80–90. https://doi.org/10.1016/j.ugj.2024.01.003 Herrmann T, Pfeiffer S (2022) Keeping the organisation in the loop: a socio-technical extension of human-centered artificial intelligence. AI Soc 38(4):1523–1542. https://doi.org/10.1007/s00146-022-01391-5 Huppert FA, So TTC (2011) Flourishing across Europe: application of a new conceptual framework for defining well-being. Soc Indic Res 110(3):837–861. https://doi.org/10.1007/s11205-011-9966-7 John YJ, Caldwell L, McCoy DE, Braganza O (2023) Dead rats, dopamine, performance metrics, and peacock tails: proxy failure is an inherent risk in goal-oriented systems. Behav Brain Sci 47:e67. https://doi.org/10.1017/s0140525x23002753 Kellogg KC, Valentine MA, Christin A (2020) Algorithms at work: the new contested terrain of control. Acad Manag Ann 14(1):366–410. https://doi.org/10.5465/annals.2018.0174 Labraña J, Ponce ER (2026) Artificial intelligence and organisational learning in universities: decision premises, paradoxes, and institutional stability. J Organ Change Manag 39:1–16. https://doi.org/10.1108/jocm-02-2025-0157 Liang W, Tadesse GA, Ho D, Fei-Fei L, Zaharia M, Zhang C, Zou J (2022) Advances, challenges and opportunities in creating data for trustworthy AI. Nat Mach Intell 4(8):669–677. https://doi.org/10.1038/s42256-022-00516-1 Lorenz W (2025) Repositioning the social mandate of social work. Nord Soc Work Res 16(1):75–87. https://doi.org/10.1080/2156857x.2025.2478838 Mac Cord D, Ramos C, De a Monteiro GF (2026) Hybrid decision architectures: exploring how facilitated AI access reconfigures startups’ organisational structures and decision-making. Int J Entrep Behav Res 32:1–33. https://doi.org/10.1108/ijebr-11-2024-1316 Martínez-Arvizu OJ, Salazar-Altamirano MA, Galván-Vela E, Anaya-Aguilar R, Anaya-Aguilar C (2025) Happiness at work in small and medium-sized enterprises: an analysis of innovation and creativity. BMC Psychol 13(1):686. https://doi.org/10.1186/s40359-025-02980-x Naser MZ (2025) A guide to machine learning epistemic ignorance, hidden paradoxes, and other tensions. Wiley Interdiscip Rev Data Min Knowl Discov 15(3):e1533. https://doi.org/10.1002/widm.70038 Parasuraman R, Riley V (1997) Humans and automation: use, misuse, disuse, abuse. Hum Factors 39(2):230–253. https://doi.org/10.1518/001872097778543886 Parycek P, Schmid V, Novak A (2023) Artificial intelligence (AI) and automation in administrative procedures: potentials, limitations, and framework conditions. J Knowl Econ 15(2):8390–8415. https://doi.org/10.1007/s13132-023-01433-3 Peeters R (2020) The agency of algorithms: understanding human–algorithm interaction in administrative decision-making. Inf Polity 25(4):507–522. https://doi.org/10.3233/ip-200253 Pirson M (2017) Humanistic management: protecting dignity and promoting well-being. Cambridge University Press Rangel-Lyne L, Salazar-Altamirano MA (2025) Mandel-AI effect: proposed measurement of the Mandela effect induced by artificial intelligence in digital social networks. Univ Psychol 24:1–19. https://doi.org/10.11144/javeriana.upsy24.maie Ravina-Ripoll R, Galvan-Vela E, Galiano-Coronil A, Ahumada-Tello E (2024) Guest editorial: happiness management—a holy grail to be discovered by companies in the age of artificial intelligence. J Manag Dev 43(2):145–149. https://doi.org/10.1108/jmd-04-2024-513 Roumbanis L (2025) On the present-future impact of AI technologies on personnel selection and the exponential increase in meta-algorithmic judgments. Futures 166:103538. https://doi.org/10.1016/j.futures.2025.103538 Ryan RM, Deci EL (2001) On happiness and human potentials: a review of research on hedonic and eudaimonic well-being. Annu Rev Psychol 52(1):141–166. https://doi.org/10.1146/annurev.psych.52.1.141 Ryff CD (1989) Happiness is everything, or is it? Explorations on the meaning of psychological well-being. J Personal Soc Psychol 57(6):1069–1081. https://doi.org/10.1037/0022-3514.57.6.1069 Salazar-Altamirano MA, Galván-Vela E, Ravina-Ripoll R, Martínez-Arvizu OJ (2025) Design and validation of an instrument for evaluating workplace happiness. J Manag Organ 32(1):21–54. https://doi.org/10.1017/jmo.2025.10071 Sergeeva AV, Leonardi PM, Faraj S (2026) Beyond the tool view of AI: intelligent technologies and the emergence of new epistemic regimes. Strateg Organ 24(2):463–478. https://doi.org/10.1177/14761270261436479 Simon HA (1947) A comment on “The science of public administration.” Public Adm Rev 7(3):200. https://doi.org/10.2307/972716 Sonnentag S, Tay L, Shoshan HN (2023) A review on health and well-being at work: more than stressors and strains. Pers Psychol 76(2):473–510. https://doi.org/10.1111/peps.12572 Spanuth A, Urbano D (2023) Exploring social enterprise legitimacy within ecosystems from an institutional approach: a systematic literature review and research agenda. Int J Manag Rev 26(2):211–231. https://doi.org/10.1111/ijmr.12349 Velotto G (2025) The meta-layered framework: a diagnostic approach to ethical pluralism in human–AI systems—paradox, pluralism, and the moral architecture of adaptive intelligence. AI Ethics 6(1):51. https://doi.org/10.1007/s43681-025-00880-9 Yusif S, Hafeez-Baig A (2024) Impact of stakeholder engagement strategies on managerial cognitive decision-making: the context of CSP and CSR. Soc Responsib J 20(6):1101–1121. https://doi.org/10.1108/srj-05-2023-0295 Zhang Z, Wang J, Jia M (2021) Multilevel examination of how and when socially responsible Human Resource Management improves the well-being of employees. J Bus Ethics 176(1):55–71. https://doi.org/10.1007/s10551-020-04700-4 Funding Funding for open access publishing: Universidad de Cádiz/CBUA. Author information Authors and Affiliations Contributions M.A.S.-A. and R.R.-R. jointly conceived the study and developed the core theoretical framework. M.A.S.-A. led the literature analysis, conceptual development, manuscript drafting, and preparation of tables and figures. R.R.-R. contributed to the theoretical refinement, critical revision of the intellectual content, and strategic positioning of the manuscript within the fields of artificial intelligence, organisational studies, and well-being research. Both authors reviewed, edited, and approved the final version of the manuscript and agree to be accountable for all aspects of the work. Corresponding authors Ethics declarations Conflict of interest The authors declare no competing interests. Generative AI and AI-assisted technologies in the manuscript preparation process During the preparation of this work, the authors used AI-assisted tools exclusively for grammar checking and translation support. The authors fully developed all substantive content, theoretical development, arguments and interpretations. The manuscript was critically reviewed and approved by the authors, who take full responsibility for its content. Additional information Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Rights and permissions Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. About this article Cite this article Salazar-Altamirano, M.A., Ravina-Ripoll, R. Beyond efficiency: epistemic well-being at work under algorithmic governance. AI & Soc (2026). https://doi.org/10.1007/s00146-026-03351-9 Received: Accepted: Published: Version of record: DOI: https://doi.org/10.1007/s00146-026-03351-9