AI-mediated bodies and the spatial production of embodied life: Rethinking sport,fitness and health in human geography

Introduction

Artificial intelligence is increasingly embedded in the environments through which bodies move, train, recover, compete and are cared for. In elite sport, algorithmic systems monitor performance, estimate fatigue, predict injury risk and guide tactical, selection and medical decisions. In commercial fitness, wearables, apps, smart devices and computer-vision systems shape exercise routines, recovery habits and norms of progress. In digital health, connected sensors, remote monitoring systems and predictive dashboards reorganize how bodily states are observed, interpreted and acted upon. These developments are often framed through innovation, personalization, optimization, surveillance or ethics. Those framings matter, but they do not fully capture what is changing. The expansion of AI-mediated bodily practice is also a geographical reorganization of the environments through which embodied life becomes knowable, comparable, actionable and governable (Longhurst, 1997).

This article uses ‘AI’ and ‘intelligent systems’ in a deliberately practical rather than narrowly technical sense. The terms refer to data-intensive computational systems that classify bodily traces, generate predictions or recommendations and help organize practical intervention. This includes machine-learning models, predictive dashboards, app-based recommendation systems, remote monitoring platforms and computer-vision tools. The article therefore does not reduce AI-mediated embodiment to automation. Many of the systems discussed here do not replace human decision-making. Instead, they reorganize the environments in which decisions are made by making some bodily signals, baselines, risks and futures more visible, authoritative and actionable than others.

A substantial body of work already examines AI, datafication and digital mediation in sport, fitness and health, but these discussions remain distributed across several research fields. In sport studies and critical sport scholarship, research has examined performance analytics, athlete monitoring, biometric data, surveillance and optimization in relation to training, injury management, performance enhancement and athlete governance (Karkazis and Fishman, 2017; Millington and Millington, 2015). In self-tracking scholarship and digital health research, attention has focused on metrics, dashboards, behavioural nudges, remote monitoring and norms of continuous self-management (Ajana, 2017; Lupton, 2016; Shaw et al., 2019). In science and technology studies, critical data studies and AI ethics, scholars have examined classification, prediction, automation, infrastructural power, algorithmic bias, accountability and the ethics of data-intensive decision-making (Beer, 2017; Mittelstadt et al., 2016). Work in digital geographies has shown how software, platforms and data-intensive systems reshape spatial relations, everyday life and governance (Ash et al., 2018; Leszczynski, 2020; Maalsen, 2023, 2024, 2025). Health geography and geographies of care have further shown that bodies, care and domestic environments are spatially organized rather than merely clinically managed (Milligan and Wiles, 2010; Parr, 2002). Yet these conversations have rarely been brought together around a more specific geographical question: how are algorithmic systems helping to produce the spatial conditions under which bodies are interpreted, adjusted and governed (Colls, 2012; Osborne and Jones, 2022)?

This article argues that the key limitation of existing work is not that it overlooks algorithmic power, but that it too often treats embodied space as the setting of AI rather than one of its effects. AI is commonly treated as a tool, layer or application operating within already formed environments of bodily practice: the training centre, the home, the clinic, the gym, the running route. Even critical accounts often retain this architecture by asking whether particular systems are fair, transparent, trustworthy or ethically justified. These questions are necessary, but they often presume a stable body and a stable environment onto which algorithmic systems are later added. A geographical approach requires a different starting point (Parr, 2002; Pykett, 2018). The issue is not only that bodies are increasingly measured through AI-mediated systems. It is that the environments in which bodily judgment occurs are themselves being reorganized through extraction, classification, prediction and intervention. This means that the article’s contribution is not simply to bring AI-mediated bodies into human geography but to show how spatial analysis changes what can be seen in wider debates on AI, embodiment and care.

AI-mediated embodiment is not best understood as the application of intelligent systems to pre-existing bodily environments, but as part of the spatial production of embodied life itself. The central question is therefore no longer simply what AI can do for sport, fitness or health, nor only whether such systems improve performance, increase adherence or enhance prevention. More fundamentally, the question becomes how algorithmic systems participate in producing the settings through which embodied life is sensed, known, compared, adjusted and governed. Sport, fitness and health then appear not as isolated sectors of technological innovation but as connected terrains through which a broader reorganization of bodily environments becomes visible. The implication is that AI-mediated embodiment is not only a technical or ethical problem, but a transformation in how bodies, spaces and futures are made mutually actionable. In practical terms, training centres, homes, clinics, gyms and everyday routes become environments in which bodily data are extracted, future states are anticipated, conduct is adjusted and responsibility is redistributed.

This shift also reveals a redistribution of bodily authority. By bodily authority, I mean the practical power to define what a body’s condition means, which bodily signals matter, what counts as risk, readiness, recovery or deterioration and whose judgment is treated as legitimate when embodied experience, professional expertise and algorithmic recommendation diverge. Bodily authority is therefore not reducible to formal decision rights. Coaches may remain formally responsible for training decisions, clinicians for treatment choices and users for whether to follow recommendations. Yet they increasingly act within environments where certain futures are made more visible, certain baselines more authoritative and certain departures more difficult to justify than others. In practical terms, this means that judgments about training load, recovery, readiness, adherence, deterioration and care may increasingly have to be defended in relation to system-generated baselines and projected futures. The issue is not simply whether humans remain ‘in the loop’, but how the loop itself is spatially organized, and how algorithmic environments help define what counts as prudent, responsible or negligent action.

To capture this transformation, the article develops the concept of algorithmically produced embodied space. The concept refers not simply to digitized embodiment or quantified selfhood, but to environments in which bodily life is reorganized through data extraction, normative classification, anticipatory calculation and practical intervention. These processes do not merely act upon bodies already situated in space. They help constitute the spatial conditions under which bodies become visible, comparable, adjustable and governable. The point is not that human judgment disappears, but that AI-mediated environments redistribute embodied authority without displacing formal responsibility.

This is a geographical problem because environments are not neutral containers but socially, materially and politically produced conditions of life (Colls, 2012; Longhurst, 1997). If that insight is taken seriously, AI in bodily domains cannot be approached simply as a set of tools acting within pre-existing spaces. The more pressing question is how training centres become anticipatory risk environments, how homes become nodes of platformized wellness or care, how gyms and routes become spaces of metricized comparison and behavioural coordination and how these transformations are distributed unevenly across bodies and places.

This intervention also builds on, but departs from, existing geographical work on sport, fitness, health and wellbeing. Geographies of sport have examined how sporting practices are embedded in place, landscape, infrastructure, mobility, identity and unequal access to sporting spaces (Bale, 2003; Vertinsky and Bale, 2004). Work on fitness, physical activity and wellbeing has shown how exercise, running, lifestyle routines and affective atmospheres are organized through everyday environments rather than individual choice alone (Coen et al., 2020; Cook and Larsen, 2022). Health geography and geographies of care have demonstrated that illness, care, recovery and bodily vulnerability are distributed across homes, clinics, communities and wider infrastructures (Milligan and Wiles, 2010; Parr, 2002). These literature studies are important because they already show that embodied practices are spatially situated. However, AI-mediated embodiment requires a further step: the issue is not only that sport, fitness and health occur in space, but that algorithmic systems increasingly help produce the spatial conditions under which bodily states are sensed, classified, anticipated and acted upon.

The article makes two linked interventions. First, it argues that much existing discussion of AI in bodily domains under-theorizes the spatial production of embodied life by treating bodily environments as prior to the systems now reorganizing them. Second, it proposes the concept of algorithmically produced embodied space to theorize how bodily environments are constituted through linked processes of extraction, classification, anticipation and intervention. The significance of this concept lies in showing that AI-mediated environments reshape not only how bodies are known but also how bodily futures are acted upon, how authority is distributed and how inequalities of legibility and care are produced. These dynamics matter in concrete settings where athletes negotiate return-to-play, fitness users interpret platform scores and patients or caregivers respond to remote monitoring alerts. The aim is not to collapse sport, fitness and health into a single field, but to read them as connected terrains through which a broader geographical transformation can be analysed. Elite sport provides a particularly intense anchor because predictive monitoring, risk management and algorithmic judgment are highly concentrated there. Commercial fitness shows how similar logics diffuse into ordinary bodily routine. Digital health reveals how these processes become entangled with care, triage and domestic responsibility (Milligan and Wiles, 2010).

The stakes differ across these terrains. In elite sport, the problem is not simply whether predictive systems enhance performance, but how they reorganize decisions about workload, injury risk, readiness and return-to-play. Athletes are most directly affected because felt readiness, pain, fatigue and willingness to compete may be reinterpreted through system-generated risk profiles; coaches and medical staff are also affected because their judgments become increasingly accountable to predictive baselines. In commercial fitness, users may gain motivation and self-knowledge, but they are also enrolled into platform environments where effort, consistency, progress and failure are continuously classified, compared and monetized. In digital health, remote monitoring may support earlier intervention, but it can also shift vigilance, anxiety and responsibility into domestic spaces, especially for patients and caregivers already unequally positioned in relation to care infrastructures.

The article proceeds in six sections. The next section develops the move from AI-as-application to the spatial production of embodied life. The following section elaborates the concept of algorithmically produced embodied space and distinguishes it from adjacent work in digital geographies, self-tracking studies and platform research. The third substantive section treats sport, fitness and health as connected terrains of algorithmic embodiment. The fourth examines uneven geographies of access, legibility and valuation. The fifth develops an account of anticipatory environments and bodily authority. The final substantive section outlines a human-geographical agenda for research on AI-mediated embodiment before the article concludes.

From AI applications to the spatial production of embodied life

A large share of current discussion treats AI in bodily domains as a matter of application. In this framing, the central question is what AI can do for sport, fitness or health: whether it can improve prediction, personalize intervention, enhance prevention or support decision-making. Even critical accounts often retain this architecture by asking whether such applications are fair, transparent, trustworthy or ethically acceptable. This work has generated valuable insights, especially around bias, surveillance, implementation and responsibility. Yet it also rests on a limiting assumption: that bodily environments are already constituted, and that algorithmic systems subsequently enter them as external aids or risks. A geographical approach requires a different starting point. The more fundamental question is not what AI does within already formed spaces of bodily practice, but how such spaces are being reorganized through algorithmic mediation.

This limitation is visible across several adjacent literatures. Research on self-tracking and quantified life has shown how digital devices encourage subjects to view bodily practice through metrics, dashboards and norms of continuous improvement (Ajana, 2017; Lupton, 2016; Ruckenstein and Pantzar, 2017). This work has been crucial for understanding responsibilization, datafication and the pedagogies of self-management that accompany wearable and app-based systems. Yet in much of this literature, space appears mainly as background: the home, the gym, the street or the clinic as pre-given settings in which tracking takes place. What remains less developed is how these settings are themselves remade when bodily knowledge, comparison and adjustment become increasingly dependent on connected, predictive and platformized systems (Osborne and Jones, 2022; Williamson, 2015).

A similar narrowing appears in work on digital care. Research on telehealth, remote monitoring and digital therapeutics often asks whether care can be delivered more efficiently, more accessibly or more responsively through connected devices and algorithmic systems (Shaw et al., 2019). Those are important questions. However, the framing can obscure a broader transformation. Care is not simply extended into the home or distributed across existing settings. Domestic environments are being reorganized as spaces of continuous physiological visibility, triage and behavioural adjustment (Milligan and Wiles, 2010; Petrakaki et al., 2021). The home does not merely receive digital care; it is increasingly remade as a site in which software-mediated judgment becomes routine (Maslen, 2017; Parr, 2002).

The same point applies to sport. Discussions of performance analytics, athlete monitoring and predictive modelling often focus on whether systems improve performance, reduce injury risk or support better decisions. Critical responses ask whether such systems intensify surveillance, erode autonomy or generate unfair competitive advantages. These debates matter, but they often leave the spatial setting of bodily judgment under-theorized. Training centres, rehabilitation settings and performance environments are not simply sites where more information is available. They are increasingly organized around ongoing capture, real-time comparison, baseline stabilization and anticipatory adjustment. The relevant transformation is not only informational. It is environmental.

Digital geographies have long unsettled the assumption that software acts upon pre-existing space from the outside. Work on code/space, software-sorted geographies and the automatic production of space has shown that code and software do not simply operate in space; they participate in producing it (Dodge and Kitchin, 2005; Graham, 2005; Kitchin and Dodge, 2011; Thrift and French, 2002). More recent work has extended this insight through platformization, algorithmic epistemologies and AI-inflected digital geographies (Ash et al., 2018; Leszczynski, 2020; Maalsen, 2023, 2024, 2025). Yet the body has not always been placed at the centre of these transformations. Where bodily domains are discussed, they often appear as examples of datafication, monitoring or governance rather than as a primary terrain through which the spatial reorganization of embodied life becomes especially visible (Osborne and Jones, 2022).

The problem, then, is not that existing work on self-tracking, digital care, sport analytics or AI ethics is misguided. It is that these framings remain too narrow when they approach sport, fitness and health primarily as domains of application. What disappears in this view is the environmental shift through which homes, clinics, training centres, gyms and everyday routes are remade as spaces of ongoing extraction, comparison, anticipation and intervention. If AI is approached merely as a tool acting on already formed bodily spaces, then the transformation of those spaces themselves recedes from view.

This matters especially for human geography. Geography has long emphasized that environments are socially and materially produced rather than given. It has also shown that spatial arrangements are inseparable from authority, access, inequality and the organization of everyday life. Taken together, these insights suggest that AI-mediated bodily practice cannot be treated simply as a question of use, impact or adoption. The more pertinent question is how algorithmic systems become part of the environments through which bodily life is organized. How do training centres become anticipatory risk environments? How do homes become sites of platformized wellness or distributed care? How do routine routes and ordinary exercise spaces become environments of metricized comparison and behavioural guidance? These are central to understanding the changing geographies of embodiment.

The move from application to spatial production also clarifies why the issue cannot be resolved by pointing out that humans remain involved. Formal decision rights do not settle where practical authority lies. A coach may still decide whether to increase training load, a clinician whether to escalate treatment and a user whether to follow a recommendation. Yet those decisions are increasingly made within environments where algorithmic outputs help define salient baselines, projected risks and prudent actions. The issue is therefore not only whether AI assists bodily judgment but also how it reorganizes the settings in which judgment takes place.

What is needed, then, is a conceptual language capable of treating AI-mediated bodily life not merely as a question of tools, applications or impacts but as a reorganization of embodied environments. The next section develops such a language through the concept of algorithmically produced embodied space.

Algorithmically produced embodied space

The concept of algorithmically produced embodied space is proposed here to capture a transformation more specific than digitized embodiment, quantified selfhood or platformized practice. It refers to environments in which bodily life is reorganized through a linked sequence of extraction, classification, anticipation and intervention. These processes do not simply act upon bodies already situated in space. They help constitute the spatial conditions under which bodies become visible, comparable, adjustable and governable.

The article uses ‘bodies’ in the plural deliberately. Bodies are not treated here as generic biological objects or neutral data sources, but as material, sensing, vulnerable and socially inscribed entities whose capacities, risks and meanings are shaped through relations with environments, institutions and technologies. The relevant body is therefore both fleshy and semiotic: it moves, tires, hurts and recovers, but it is also classified through gendered, racialized, ableist, clinical, athletic and commercial norms. Algorithmically produced embodied space matters because it does not simply measure such bodies. It reorganizes the conditions under which sensations, capacities, vulnerabilities and futures become interpretable and actionable.

This formulation matters because it shifts analysis away from the idea that AI merely operates within bodily environments. Instead, it foregrounds how algorithmic systems help produce such environments as practical settings of knowledge and action. Bodies are rendered extractable through sensors, cameras, interfaces, wearables, connected devices, locative traces and behavioural histories (Osborne et al., 2023; Osborne and Jones, 2017). Extracted traces are stabilized into baselines, thresholds, categories and risk profiles through classificatory schemes. These classifications support anticipatory calculations through which possible futures are made actionable in the present. Finally, prediction is folded back into embodied conduct through prompts, alerts, frictions, recommendations and decisions. What emerges is not simply a more data-rich environment, but an embodied space reorganized around differential visibility, normative evaluation and future-oriented intervention.

The first moment in this sequence is extraction. Bodily life becomes continuously available to capture through sensors, cameras, biometric devices, smart equipment, touch interfaces and mobile platforms. Heart rate, movement quality, sleep, route choice, effort, cadence, readiness, recovery, adherence and physiological fluctuation are rendered extractable and storable. For example, a runner’s heart rate, cadence and route may be captured through a wearable device; an athlete’s acceleration and deceleration patterns may be captured through GPS; and a patient’s blood pressure or glucose readings may be transmitted from home to a clinical dashboard. Extraction matters spatially because it does not occur evenly. Some environments become dense with capture infrastructures, while others remain comparatively opaque. Some bodies become continuously available to analysis, while others are visible only intermittently or under conditions of breakdown. Extraction is therefore not merely technical acquisition. It is the beginning of a politics of bodily visibility in space.

The second moment is classification. Once extracted, bodily traces are sorted into norms, baselines, thresholds, anomalies and profiles of performance, fatigue, compliance, risk or concern. Classification stabilizes assumptions about what counts as progress, readiness, decline, deviation or failure. For example, extracted traces may be classified as normal recovery, insufficient readiness, elevated injury risk, poor adherence or deviation from a population-based threshold. These assumptions often appear as neutral outputs of system processing, yet they are inseparable from prior design decisions, training datasets and institutional judgments. Classification matters geographically because it organizes environments normatively. It establishes what is ordinary, what is excessive, what is tolerable and what requires intervention.

The third moment is anticipation. AI-mediated systems increasingly do not just record what a body has done; they estimate what may happen next. Injury probabilities, recovery windows, relapse forecasts, adherence risks and deterioration alerts render future bodily states calculable and actionable. For example, a workload model may forecast reinjury risk, a fitness platform may predict declining readiness or a remote monitoring system may flag possible deterioration before symptoms become clinically acute. This future orientation is not only epistemic. It is spatial and political. It changes what kinds of futures are made present, what kinds of action appear prudent and what forms of uncertainty remain acceptable. Bodily life is governed not only through present states or retrospective data but also through the anticipation of possible states not yet realized.

The fourth moment is intervention. Extraction, classification and anticipation matter because they are connected to practical modulation. Systems generate altered workloads, route adjustments, recovery suggestions, behavioural prompts, triage alerts, friction points, care escalations and decision thresholds. For example, systems may recommend rest, reduce training load, prompt a user to exercise, escalate a patient for review or mark a body as requiring closer supervision. Intervention closes the loop between representation and action. The environment becomes not simply a site in which bodily data are observed, but a medium through which conduct is continuously shaped.

Taken together, these four moments describe a process through which embodied environments are produced rather than merely instrumented. A training centre saturated with dashboards and predictive monitoring, a home organized through fitness subscriptions and recovery platforms or a domestic care environment linked to remote sensing and triage systems are not simply old spaces with new devices added on. They are reconstituted settings in which bodily life is rendered differentially visible, normatively assessed and repeatedly adjusted. The concept of algorithmically produced embodied space names that environmental transformation.

The concept is not intended to displace adjacent work in digital geographies, code/space scholarship, self-tracking studies or platform research. Rather, it addresses a limitation produced when these literature studies remain only partially connected. Digital geographies and code/space scholarship explain how software becomes constitutive of environments, but they do not always foreground the body as the terrain through which such reorganization becomes lived, judged and adjusted (Dodge and Kitchin, 2005; Kitchin and Dodge, 2011; Thrift and French, 2002). Self-tracking scholarship has shown how everyday life is datafied through responsibilization and norms of self-improvement, but it often treats space as a background setting for measurement rather than an environment being actively remade (Ajana, 2017; Lupton, 2016). Platform studies illuminate infrastructural dependency and commercial mediation, yet they say less about how algorithmic systems become practical baselines for interpreting bodily deviation within embodied settings (Plantin et al., 2018; van Dijck et al., 2018). The concept of algorithmically produced embodied space is needed because the contemporary transformation at stake involves all three dimensions at once: environmental reorganization, anticipatory governance and the redistribution of embodied authority (Osborne and Jones, 2022).

Its added value lies especially in clarifying the question of bodily authority. Authority here should not be reduced to formal decision rights. The more consequential issue is where practical interpretive authority sits within a given environment. Who defines the relevant baseline? Who names risk? What counts as prudent deviation? Which forms of embodied judgment remain legitimate in the face of predictive recommendation? AI-mediated systems can redistribute such authority even when humans formally remain responsible for outcomes. Coaches, clinicians and users may continue to decide, but they increasingly do so within settings where algorithmic outputs help structure what appears sensible, defensible or negligent.

The concept also helps distinguish algorithmically produced embodied space from broader claims about digitization. Digitization alone does not explain how bodily environments become future-oriented settings of guidance and adjustment. Nor does platformization alone explain how normative thresholds and predictive baselines become embedded in practical routines of bodily conduct. Algorithmically produced embodied space refers to a more specific transformation: the production of embodied environments in which extraction, classification, anticipation and intervention are linked tightly enough that bodies are governed through continuous relations among data, prediction and action.

The next section develops this argument empirically by treating sport, fitness and health not as parallel sectors but as connected terrains through which the spatial production of AI-mediated embodiment becomes particularly visible.

Sport, fitness and health as connected terrains of algorithmic embodiment

Sport, fitness and health are often treated as distinct sectors governed by different institutional norms, professional logics and forms of expertise. Yet they are increasingly connected through shared infrastructures of sensing, tracking, prediction and behavioural adjustment. Rather than reviewing them as parallel literature, this article treats them as linked terrains through which a wider geographical transformation becomes visible. Elite sport provides a particularly intense and analytically clarifying anchor because predictive monitoring, risk management and algorithmic judgment are highly concentrated there. Commercial fitness shows how related logics diffuse into ordinary bodily routine. Digital health reveals how these processes become entangled with care, triage and domestic responsibility. Taken together, these terrains show how AI-mediated embodiment is not a sectoral anomaly but an increasingly general mode of spatially organized bodily governance.

Across these terrains, the central problems are not identical, but they share a common spatial form. In elite sport, predictive systems affect how injury, fatigue and return are judged within high-pressure institutions, where risk decisions involve athletes, clinicians and coaches and may carry consequences for career trajectories, performance opportunities and bodily harm. In commercial fitness, tracking platforms shape how ordinary users understand effort, progress and failure within commercial interface environments, creating possibilities for motivation and self-knowledge while also intensifying self-surveillance, comparison and platform dependency. In digital health, remote monitoring changes how patients and caregivers manage risk, deterioration and responsibility within domestic space, potentially extending care while shifting vigilance and response burdens into everyday life. These problems are unevenly distributed: some bodies become more supported and actionable, while others become more exposed to surveillance, misclassification, self-blame or intensified care burdens.

Elite sport offers the clearest view of algorithmically produced embodied space because it combines high stakes, dense sensing infrastructures and strong pressures toward anticipatory action. Training centres are increasingly organized through GPS systems, heart-rate monitoring, biomechanical tracking, recovery dashboards and injury-prediction tools that render bodily states legible as dynamic patterns of readiness, deviation and risk. What matters here is not simply that more information is available. The training environment itself is reorganized around continuous capture, comparison and adjustment. Athletes move within spaces where workload, fatigue, vulnerability and return-to-play timing are made actionable through predictive systems, while coaches and medical staff make judgments in settings where algorithmic outputs increasingly shape what counts as prudent load, acceptable uncertainty and defensible intervention. Bodily authority is therefore not eliminated, but redistributed.

Sport is analytically useful not because it is representative in every respect, but because it magnifies processes that are otherwise easier to miss. Predictive systems in elite sport make visible a broader environmental shift: algorithmic environments do not simply supply additional information; they redefine the practical space within which bodily judgment occurs. Training centres, rehabilitation sites and performance settings become anticipatory environments in which action is increasingly organized around projected futures.

A particularly revealing example is the return-to-play setting. Return-to-sport decisions are already complex, recurrent and consequential forms of risk management involving athletes, clinicians and coaches, rather than isolated clinical decisions made only at the end of recovery (Ardern et al., 2016). AI-mediated monitoring intensifies this setting because readiness scores, workload histories, biomechanical markers and reinjury-risk estimates can make projected vulnerability actionable before full competitive return. This may support caution and improve decision-making, especially where training load and injury risk are already understood as dynamically related (Gabbett, 2016). Yet it also changes how bodily authority is distributed. Athletes may find felt readiness, pain, fatigue or willingness to compete reinterpreted through risk models; clinicians may have to defend return decisions in relation to algorithmic baselines; coaches may find reintegration constrained by projected vulnerability. Return-to-play environments therefore show with particular clarity that AI-mediated embodiment is not just about better information. It is about the production of settings in which uncertainty is reorganized, prudence is recalibrated and deviation from predictive guidance becomes more difficult to justify.

This is why sport serves here as a primary anchor rather than merely one case among others. It reveals in concentrated form how bodily spaces are reconstituted through extraction, classification, anticipation and intervention.

Commercial fitness demonstrates how related dynamics diffuse beyond specialized institutions into the ordinary spaces of everyday life. Wearables, subscription platforms, app-based coaching systems, smart exercise devices and recovery services reorganize homes, neighbourhood routes and gyms as interface-dependent environments of ongoing bodily coordination. Exercise no longer simply occurs in space and is later recorded (Toner et al., 2023). It increasingly unfolds through prompts, rankings, streaks, adaptive recommendations and compliance signals that help script when, where and how movement should take place. What appears in elite sport as dense predictive monitoring reappears here in lower-intensity but more continuous form: as repeated nudging, platform dependency and routine adjustment built into everyday bodily environments (Williamson, 2015). These systems appear to personalize movement, yet they also stabilize particular norms of consistency, progress, readiness and sensible effort (Fotopoulou and O’Riordan, 2017). The result is not only quantified selfhood, but the platformization of ordinary embodied space.

This diffusion into everyday life matters because it changes the scale and social reach of algorithmic bodily governance. In elite sport, predictive environments are concentrated, visible and institutionally managed. In commercial fitness, similar logics are banalized through subscription ecosystems, consumer interfaces and ordinary daily routine. The evening run, the home workout, the sleep cycle and the recovery day are increasingly mediated through systems that normalize self-monitoring and repeated adjustment (Toner et al., 2023). The ordinary is therefore one of the most significant sites of algorithmic embodiment (Cai et al., 2021; Ouyang et al., 2022). Fitness platforms make clear that algorithmically produced embodied space is not confined to exceptional settings of expertise, but it is becoming woven into mundane rhythms of bodily life.

Digital health extends these transformations into spaces of care, prevention and domestic responsibility. Remote monitoring systems, connected sensors, telehealth platforms and predictive dashboards bring bodily observation and triage into homes and everyday routines. This may improve convenience, continuity and early detection, but it also reorganizes domestic environments as spaces of ongoing physiological visibility and anticipatory response (Milligan and Wiles, 2010). Patients are increasingly expected to notice trends, respond to alerts and align conduct with projected indicators before deterioration becomes clinically evident (Petrakaki et al., 2021). Care is therefore not merely delivered through digital devices in pre-existing settings. The home itself is reconstituted as a node within wider infrastructures of software-mediated judgment, institutional protocol and commercial platform coordination (Maslen, 2017). In this setting, algorithmically produced embodied space becomes visible as a distributed form of care governance.

The value of bringing sport, fitness and health together lies in showing how a broader spatial logic moves across domains that are too often analysed in isolation. Performance optimization techniques developed in elite sport migrate into consumer fitness. Practices of adherence, readiness and preventive self-management move between health, wellness and workplace settings. Remote monitoring infrastructures developed in care contexts intersect with platformized routines of self-tracking and behavioural guidance. These flows matter geographically because they reveal that AI-mediated embodiment is not reducible to any one sectoral discourse. It is part of a wider reorganization of bodily environments across institutional and everyday settings alike.

Treating these terrains together also clarifies that differences of purpose do not erase common spatial forms. Elite sport is highly competitive and professionally managed. Commercial fitness is market-oriented and interface-centred. Health remains entangled with care, prevention and formal responsibility. Yet across all three, bodily life is increasingly organized through environments built for extraction, normative sorting, future-oriented calculation and practical intervention. In each case, algorithmic systems are not simply appended to bodily practice. They help define what becomes visible, what counts as deviation, which futures become actionable and how authority is distributed.

Uneven geographies of access, legibility and valuation

Algorithmically produced embodied space is not evenly distributed. It is differentiated by unequal access to predictive infrastructures, unequal fit within classificatory systems and unequal conversion of bodily data into social, institutional and economic value. Any account that focuses only on innovation, personalization or efficiency will miss the extent to which AI-mediated embodiment is also a field of uneven visibility, differential governability and stratified advantage. A geographical approach is essential because the benefits, burdens and forms of authority generated by algorithmic bodily environments are always spatially distributed rather than universally shared.

The first unevenness concerns access. Entry into predictive bodily infrastructures depends on far more than the presence or absence of devices. It requires connectivity, interoperability, interpretive support, institutional capacity and often ongoing financial commitment. Elite clubs, specialized clinics, affluent households and well-resourced organizations are more likely to access advanced sensing systems, integrated dashboards and expert interpretation. Others rely on generic consumer tools, fragmented platforms or no meaningful support at all. A training environment with dense monitoring infrastructure and dedicated staff does not generate the same possibilities as a community setting mediated through a basic app. Likewise, a household linked to remote care services, connected devices and platform support is not equivalent to one without such infrastructures. Spatial inequality thus becomes inequality in foresight, guidance and governability.

This matters because access is not merely about participation in technological modernity. It shapes which bodies and places can be rendered actionable through predictive systems and which remain only partially visible. Some environments are saturated with capture technologies and interpretive support, while others are drawn into data extraction without comparable resources for contextual judgment or meaningful care.

The second unevenness concerns legibility. Bodies do not enter algorithmic systems on equal terms. AI-mediated environments are built through training data, design choices, classificatory norms and assumptions about desirable trajectories. These assumptions may privilege some bodily forms, capacities and temporal routines while misreading, flattening or marginalizing others. Gendered, racialized, disabled, ageing or otherwise non-standardized bodies may be forced into ill-fitting baselines or treated as anomalous when they diverge from system expectations (Biruk, 2024). In sport, assumptions about optimal workload, recovery or readiness may be derived from particular populations and generalized beyond them. In health, predictive tools may reproduce earlier clinical inequities. In fitness, routine metrics of improvement may privilege specific bodily ideals and lifestyles. Legibility is therefore not simply a matter of technical accuracy. It is a political-spatial question about which bodies can be computationally known on acceptable terms and which remain partially misrecognized within dominant systems of calculation.

This matters because legibility is often mistaken for neutrality. When algorithmic systems render some bodies more clearly than others, the result is not just epistemic difference but differentiated exposure to intervention. Bodies that fit prevailing classificatory schemes may be more easily guided, optimized or supported. Bodies that do not fit may be under-served, over-corrected or repeatedly treated as noise, anomaly or risk.

The third unevenness concerns valuation. AI-mediated environments do not simply reveal bodily worth; they help produce it through intersecting value systems. Economically, bodily data may generate value through subscriptions, data extraction, engagement, retention, targeted services and platform lock-in (Plantin et al., 2018; van Dijck et al., 2018). Institutionally, bodies that are highly measurable, comparable and optimizable may become more valuable because they are easier to manage, discipline, select, insure, treat or compare. In governance terms, some bodies become valuable because they can be rendered actionable through predictive systems, while others become costly, uncertain or marginal because they do not align easily with dominant infrastructures. Valuation is therefore not only symbolic recognition. It is a process through which bodies are differentially incorporated into commercial, institutional and governing circuits of value.

This process is visible across sport, fitness and health. An athlete may benefit from refined monitoring while also becoming more valuable as a measurable performance asset and more exposed to intensified discipline. A fitness user may receive personalized guidance while simultaneously generating behavioural data that sustain engagement, subscription revenue and platform development. A patient may gain earlier warning while also becoming responsible for producing usable data and responding appropriately to system alerts. Conversely, bodies that are harder to classify, less profitable to support, more costly to interpret or less compatible with standardized thresholds may be rendered less valuable, even when they are more in need of care. These are not contradictions external to the system. They are constitutive of how algorithmic bodily environments translate visibility into economic, institutional and governing value.

These three unevennesses are analytically distinct but practically entangled. Not all bodies and places can enter predictive infrastructures on equal terms. Not all bodies are read equally well once inside them. And not all bodies benefit equally from being rendered legible through such systems. This is why bias and exclusion cannot be treated simply as technical faults awaiting better design. They are spatially organized features of contemporary data-intensive environments. Some places are rich in sensing, support and anticipatory capacity. Others are poor in all three. Some bodies are calibrated as central to system design, while others remain marginal, noisy or over-exposed to correction.

This perspective also complicates celebratory claims about democratization. Wearables, apps and connected health services can indeed broaden access to some forms of bodily knowledge and coordination. But access to metrics is not the same as access to meaningful interpretation, supportive institutions or equitable care. Low-cost systems may extend monitoring while narrowing interpretive depth. Ubiquitous tracking may create a sense of empowerment while intensifying responsibilization and dependence on opaque infrastructures. The spread of AI-mediated embodiment can therefore widen participation and deepen inequality at the same time.

Anticipatory environments and bodily authority

A defining feature of algorithmically produced embodied space is its anticipatory organization. These environments do not simply describe present bodily states; they make possible futures actionable in the present. Injury probabilities, readiness estimates, recovery windows, relapse forecasts, adherence risks and deterioration alerts render projected futures into ordinary objects of practical reasoning. AI-mediated embodiment is anticipatory not simply because futures are estimated, but because environments are increasingly organized so that acting on projected futures appears prudent, normal and responsible. The significance of this shift lies not only in the growing prevalence of prediction but also in the way projected futures become embedded in the routine spatial conditions of bodily life (Amoore, 2013; Anderson, 2010). What is at stake is not only how bodies are measured but also how predictive environments redistribute bodily authority by defining which futures become actionable, which baselines become authoritative and which deviations become difficult to justify.

This anticipatory organization changes the terms of embodied action. Actors are increasingly encouraged to respond not only to what is presently felt, directly observed or clinically established, but also to system-generated estimates of latent risk and probable development (Maslen, 2017). The coach adjusts workload because projected injury risk has risen. The runner rests because readiness scores have dipped. The patient responds to an alert because predictive indicators suggest possible deterioration. Bodily judgment is thereby reorganized around futures that have not yet materialized but are rendered actionable through interfaces, dashboards, scores, thresholds and alerts.

The importance of this shift is not exhausted by saying that data now inform decisions. Human actors have always made judgments with some orientation toward the future. What is changing is that possible futures are increasingly formatted, prioritized and operationalized through infrastructural systems that claim privileged access to what is likely to occur next. Anticipation becomes environmental rather than merely cognitive. It is built into the material and interface conditions through which bodies are monitored, compared and adjusted. Homes, clinics, training centres, gyms and everyday routes become spaces in which future-oriented governance is routine rather than exceptional.

This matters because anticipation redistributes bodily authority in subtle but consequential ways. Formal responsibility may remain with coaches, clinicians or users, yet practical interpretive authority increasingly resides within infrastructures that claim privileged access to latent states and probable futures. The key issue is not whether humans remain ‘in the loop’, but how the loop itself is spatially organized. In AI-mediated environments, forecasts, baselines, thresholds and alerts help define what counts as sensible action, acceptable risk and negligent deviation. Human actors may still decide, but they increasingly do so within settings where algorithmic outputs structure the field of defensible judgment.

Bodily authority should therefore not be reduced to the nominal right to decide. A more demanding question is where interpretive authority sits within a given environment. Who defines the relevant baseline? Who names risk? What counts as prudent deviation from system output? Under what conditions does embodied feeling, professional experience or situated judgment remain legitimate when it conflicts with predictive recommendation? These questions matter because algorithmic environments can narrow the space of plausible action without ever formally displacing human agents. Prediction need not become command in order to become authoritative. It may instead function as a practical baseline from which departure increasingly requires explanation.

This is one reason why AI-mediated embodiment cannot be understood only through dramatic scenarios of automation. Much governance occurs through low-intensity, repeated interventions: reminders, thresholds, warnings, friction points, rankings, prompts and soft recommendations that cumulatively define what appears normal or responsible. These interventions are often presented as supportive rather than coercive. The point is not that all prediction is domination. It is that environments organized around prediction alter the normative atmosphere in which bodily action occurs.

A geographical perspective is especially valuable here because it makes visible how anticipation is spatially organized. Projected futures do not circulate in the abstract. They are embedded in interfaces, devices, dashboards, protocols and routines that attach prediction to particular places and practices. The future is made present in the training centre through readiness scores and load recommendations, in the home through remote monitoring alerts and behavioural nudges and in the gym or on the running route through adaptive prompts and recovery metrics. These are micro-geographies of pre-emption in which bodily futures are continuously folded into present conduct.

Anticipatory organization also intensifies responsibilization. The responsible athlete tracks trends, notices deviations and acts before strain becomes injury. The responsible fitness user responds to prompts, maintains consistency and avoids preventable decline. The responsible patient notices physiological change and acts before deterioration becomes acute. Responsibility is increasingly attached not only to care of the present body but also to alignment with projected futures. This does not mean that people respond passively. They may resist, reinterpret, ignore or strategically use system outputs. But such responses now occur within settings where prediction helps define what counts as foresight, prudence and negligence.

Seen in this light, anticipation is not an additional feature of AI-mediated embodiment layered onto pre-existing bodily spaces. It is one of the mechanisms through which those spaces are produced. Environments become algorithmically embodied not only when data are extracted or classifications assigned, but when futures are made present in ways that structure conduct. This is why anticipatory governance should be treated as central rather than secondary.

A human-geographical agenda for AI-mediated embodiment

If AI-mediated embodiment is becoming a central mode of contemporary bodily governance, then human geography should treat it not as a niche technical topic but as a wider reorganization of embodied spatial life. The argument developed here suggests that future research should move beyond narrow questions of adoption, effectiveness or ethical compliance and instead ask how bodily environments are being infrastructurally assembled, unevenly distributed and anticipatorily governed. Three lines of inquiry are especially important.

First, research should examine the infrastructures of bodily governance. This means tracing how wearables, sensors, mobile apps, analytics providers, platform services, health systems, insurers, procurement chains and subscription models are linked together across scales. The point is not only to document user experience or technological uptake, but to understand how AI-mediated bodily environments are materially and institutionally made possible. Which devices, software stacks and cloud services are required to sustain predictive bodily environments (Osborne et al., 2023)? How do interoperability standards, data-sharing arrangements and commercial partnerships shape what kinds of embodiment become actionable? How do infrastructures differ across sporting institutions, domestic settings and health systems? These questions matter because bodily governance is never simply local. It is assembled through relations that connect intimate bodily routine to distant technical standards, corporate architectures and institutional protocols.

This infrastructural focus also has consequences for how geography approaches scale. A home-based recovery routine, a gym dashboard or a training-centre monitoring system may appear local and immediate, yet each may depend on translocal arrangements of software design, data processing, procurement and commercial strategy. Human geography is especially well placed to show how intimate bodily practice is entangled with wider circuits of platformization, technical standardization and economic coordination (Pykett, 2022).

Second, more attention is needed to the spatial inequalities of legibility and care. The spread of AI-mediated bodily environments raises questions not only about who gains access to systems but also about who is made interpretable, supportable and governable through them. Which bodies fit dominant classificatory schemes, and which are misread, flattened or repeatedly rendered anomalous? Which environments provide meaningful interpretive support, and which offer extraction without care? Which populations are brought into predictive infrastructures only to be monitored more intensely or made more responsible for managing their own risk? These questions connect AI-mediated embodiment to long-standing geographical concerns with uneven development, differentiated exposure to power and stratified access to care.

A stronger research agenda would therefore examine not only moments of exclusion but also gradations of partial inclusion. Some bodies may be rendered visible enough to be governed, but not visible enough to be well understood. Some households may receive monitoring infrastructures without receiving corresponding institutional support. Some sporting settings may gain predictive systems that intensify discipline without improving welfare. Geography can clarify how algorithmic embodiment is differentiated not only by access and absence, but by variable combinations of visibility, support, burden and control.

Third, human geography should analyse the politics of anticipatory authority. If AI-mediated environments increasingly organize bodily life around projected futures, then the central question is not simply whether prediction is accurate, but how predictive systems reshape the conditions of judgment. When does predictive advice remain advisory, and when does it become the practical baseline from which deviation is difficult to justify? Under what conditions can users, athletes, patients, coaches or clinicians reinterpret, negotiate or refuse system outputs? When does assistance become subtle compulsion? How are answerability and responsibility redistributed when judgment takes place inside predictive environments rather than outside them? These questions are central if AI-mediated embodiment is to be understood not only as datafication but as a transformation in the spatial organization of bodily authority.

A focus on anticipatory authority would also open more explicit dialogue between human geography and adjacent debates on automation, governance, care and expertise. These debates are taking place across science and technology studies, critical data studies, digital health research, platform studies, health sociology, sport studies and self-tracking scholarship. Human geography is not external to these conversations. Existing work in digital geographies and code/space scholarship has already shown how software and automated systems participate in the production of spatial relations (Dodge and Kitchin, 2005; Kitchin and Dodge, 2011; Thrift and French, 2002). Health geography and geographies of care have shown how care, vulnerability and responsibility are distributed across homes, clinics and everyday environments (Milligan and Wiles, 2010; Parr, 2002). Work on embodied, affective and fitness geographies has further demonstrated how running, exercise and wellbeing are organized through situated bodily practices and ordinary spaces (Cai et al., 2021; Ouyang et al., 2022; Toner et al., 2023). Building on these dialogues, the agenda proposed here asks how predictive futures become normalized in ordinary sites: the training floor, the kitchen table, the clinic interface, the recovery app, the neighbourhood route. What matters in these settings is often not overt command, but the quiet establishment of predictive norms that shape what appears prudent, responsible or negligent.

Although this article has focused on sport, fitness and health, the proposed agenda is not confined to these domains. Similar questions arise in explicitly embodied practices such as dance, theatre and performance, where movement quality, fatigue, expressiveness, timing and bodily capacity may become subject to digital capture, algorithmic assessment or predictive management. They also arise in less obviously embodied domains such as education, workplace productivity, mobility and welfare administration, where bodies are not always named as the primary object of governance but are nonetheless monitored through attendance, attention, affect, risk, productivity, capacity or compliance. The concept of algorithmically produced embodied space is therefore useful beyond explicitly bodily fields because it asks how environments make bodies actionable even when embodiment is not the declared object of intervention.

Taken together, these three lines of inquiry suggest that AI-mediated bodies should become a major object of geographical analysis. The key issue is not simply whether particular systems improve performance, increase efficiency or extend care. It is how they reorganize the environments in which bodily life is lived, interpreted and governed, and how they do so through infrastructures, inequalities and anticipatory forms of authority. Human geography has the conceptual and methodological resources to analyse this transformation, but doing so requires moving beyond the language of application and toward a fuller account of the spatial production of embodied life.

Conclusion

This article has argued that AI should not be understood simply as a set of tools applied within already formed spaces of bodily practice. Across sport, fitness and health, algorithmic systems participate in producing the spatial conditions under which embodied life becomes visible, comparable, actionable and governable. The concept of algorithmically produced embodied space was proposed to name this transformation and to show how extraction, classification, anticipation and intervention are linked in the reorganization of bodily environments.

The significance of the proposed conceptual approach lies in what it makes visible across domains. Algorithmically produced embodied space helps identify how AI-mediated environments reorganize not only bodily knowledge but also uncertainty, responsibility, agency and inequality. In elite sport, it reframes predictive monitoring and return-to-play assessment as questions not only of performance science but also of how uncertainty, responsibility and athlete agency are spatially organized; athletes may find pain, fatigue and felt readiness reinterpreted through risk models, while clinicians and coaches must increasingly justify decisions against system-generated baselines. In commercial fitness, it shows that tracking platforms do not merely encourage self-knowledge but embed users in environments where progress, effort, adherence and failure are continuously classified and often monetized; users may experience motivation and self-care alongside self-surveillance, platform dependency, comparison and intensified responsibility for bodily optimization. In digital health, it clarifies how remote monitoring can extend care while also relocating vigilance, responsibility and risk management into the home; patients and caregivers may gain earlier warning but also bear new burdens of alert-response, anxiety and continuous self-management, especially where care infrastructures are already uneven.

These effects are psychological, bodily, social and unequal at the same time. Psychologically, anticipatory environments may intensify vigilance, anxiety, self-surveillance and the pressure to remain continuously responsive to scores, alerts and risk indicators. Bodily, they reshape how fatigue, pain, readiness, recovery and vulnerability are interpreted, sometimes making felt experience secondary to system-generated baselines. Socially, they redistribute authority among athletes, users, patients, coaches, clinicians, caregivers, platforms and institutions by changing whose judgment is treated as credible or defensible. In terms of health inequity, they can make already well-supported bodies more actionable and better managed, while leaving other bodies misread, under-supported, over-monitored or burdened with new responsibilities without equivalent care. These are not only technical or ethical issues. They are spatial problems because they concern where authority sits, where responsibility is displaced, who must live with alerts, scores and risk profiles and how bodies become differently legible and governable across environments. The value of the agenda proposed here is therefore to move analysis from whether AI systems work to how they reorganize the spatial conditions under which bodily life is interpreted, acted upon and governed.

The article has made three main claims, all of which follow from a single proposition: AI-mediated embodiment is not simply situated in space but participates in producing the embodied spaces through which bodily life is governed. First, it has challenged dominant framings that treat AI in bodily domains primarily as matters of technical application, self-tracking or digital care. Second, it has proposed a concept that foregrounds how embodied environments are produced through interconnected processes of data capture, normative sorting, future-oriented calculation and practical modulation. Third, it has shown that these transformations are unevenly distributed through geographies of access, legibility and valuation, and they are politically consequential because they redistribute bodily authority through anticipatory environments.

The broader implication is that AI-mediated embodiment should become a central object of human-geographical analysis and a stronger concern for interdisciplinary research on AI, sport, fitness, health and care. The issue is not only whether algorithmic systems improve performance, care or efficiency. It is how they reorganize the environments in which bodily life is interpreted and governed, how they make some bodies more legible and more actionable than others and how they normalize forms of predictive authority within everyday settings. Under contemporary conditions, embodiment is increasingly lived through spaces that are not merely digital, but infrastructurally assembled, differentially distributed and anticipatorily organized. The value of a geographical approach, therefore, lies in showing that AI-mediated bodily futures are produced through spatial arrangements, not merely predicted by technical systems.