IOCSS Research Paper · Sports × AI Philosophy · 2026
Abstract
Artificial intelligence is rapidly becoming embedded in the fabric of competitive sport: from training algorithms and injury prediction systems to VAR officiating, performance analytics, and athlete biometric monitoring. This paper develops a philosophical framework for analysing the ethical challenges that AI integration poses for sport. We argue that sport's value rests on three foundational pillars — fair play, embodied excellence, and public trust — and that each pillar faces distinctive threats from algorithmic judgment. Drawing on Rawlsian fairness theory, phenomenological accounts of embodiment, and trust epistemology, we propose five normative principles for a responsible AI-in-sport framework: transparency, contestability, embodiment-sensitivity, distributive access, and democratic governance. We conclude that AI in sport is not merely a technical problem but a constitutive philosophical challenge that requires sport's governing bodies, athletes, fans, and societies to deliberate about what sport is ultimately for.
1. Introduction: When Judgment Becomes Algorithmic
Sport has always been a domain of human judgment. The referee decides whether the ball crossed the line. The coach decides whether to call a timeout. The athlete decides whether to pass or shoot. These judgments are fallible, of course — sometimes spectacularly so — but their fallibility has traditionally been understood as intrinsic to sport's human character. The missed call is part of the game. The biased referee is a narrative element that generates outrage, debate, and sometimes heroism.
In the second decade of the twenty-first century, this situation is changing with remarkable speed. Artificial intelligence is entering sport not as an external observer but as an active participant in the production of competitive outcomes. Video Assistant Referee (VAR) systems now adjudicate offside calls in professional football by tracking the geometry of players' bodies to sub-centimetre precision. Machine learning platforms analyse millions of data points to evaluate athlete performance, project injury risk, and recommend training load adjustments. Natural language processing systems generate real-time scouting reports and tactical analyses. Biometric wearables capture heart rate variability, sleep quality, and musculoskeletal stress with millisecond resolution. In eSports, AI agents trained through reinforcement learning now routinely defeat the world's best human players at games of incomplete information.
These developments raise philosophical questions of fundamental importance. If an AI system determines whether a player was offside, is the outcome "fair" in any meaningful sense? If an algorithm advises a coach on which athletes to develop, who bears responsibility for the human costs of that recommendation? If an athlete's body is comprehensively monitored by biometric AI systems owned by their club, what becomes of bodily autonomy and the right to privacy? If performance enhancement and injury prevention technologies are available only to wealthy clubs and national federations, does this undermine the foundations of competitive equity?
This paper addresses these questions by developing a philosophical framework for evaluating AI's role in sport. Section 2 examines the concept of fair play and the ways in which AI systems can both promote and undermine it. Section 3 analyses the philosophical significance of embodiment in sport and the risks that algorithmic mediation poses for the value of embodied excellence. Section 4 develops an account of public trust in sport and argues that AI transparency is a necessary condition for trust's preservation. Section 5 addresses the distributive justice dimensions of AI-in-sport, focusing on access inequality. Section 6 proposes five normative principles for responsible AI governance in sport. Section 7 concludes.
2. Fair Play in the Algorithmic Age
Fair play is sport's central regulative ideal. Its content is philosophically complex, encompassing procedural fairness (the rules are applied consistently to all competitors), substantive fairness (the rules are themselves just), and what we might call constitutive fairness (the competition genuinely tests the capacities it is designed to test). AI systems intervene at all three levels.
At the procedural level, AI officiating systems can, in principle, dramatically reduce inconsistency. Human referees are subject to cognitive biases, fatigue, positioning errors, and the simple fact that a match involves hundreds of events occurring simultaneously. A well-designed AI system can process all available visual information, apply rules consistently, and do so without the psychological pressures that distort human judgment. This is the strongest argument for AI officiating: if we care about procedural fairness, shouldn't we want the most accurate and consistent application of the rules we can achieve?
The philosophical complications arise quickly. First, procedural fairness in sport is not simply a matter of accurate rule application — it also depends on the rules being designed in ways that are consistent with sport's underlying purposes. The offside rule in football, for example, is a complex normative instrument designed to balance the interests of attacking and defending teams, to create tactical complexity, and to preserve the game's excitement. When VAR systems interpret offside by tracking body geometry to sub-centimetre precision, they may be applying the rule with unprecedented accuracy while simultaneously violating its spirit. A player who is "offside" by a fraction of a centimetre as measured by a pixel-tracking system may have had no tactical advantage whatsoever from their position — the very situation the offside rule was designed to prevent. Technical accuracy in rule application can thus produce constitutive unfairness.
Second, the design of AI officiating systems necessarily encodes normative choices that are typically invisible to stakeholders. What counts as a foul? Which contact is "excessive"? These determinations require training data compiled by human annotators who bring their own normative frameworks to the task. The AI system is not neutral — it embeds and amplifies the normative choices made by its designers. When these choices are made opaque by the technical complexity of machine learning models, the possibility of democratic deliberation about the rules of sport is undermined.
Third, the geographic and economic distribution of AI officiating technology creates new forms of competitive inequity. High-precision AI officiating is available at top-tier professional levels; lower leagues, youth sport, and the sport of the global majority operates without it. This creates a two-tier officiating system in which the quality of rule application correlates with wealth and geography, rather than with the requirements of fair play.
3. Embodied Excellence and the Challenge of Algorithmic Mediation
Sport matters, philosophically speaking, because it provides a distinctive kind of human value: the cultivation and public expression of embodied excellence. The philosopher Bernard Suits argued that sport involves the voluntary adoption of unnecessary obstacles — rules that make achievement more difficult — in order to make possible a valuable activity that couldn't otherwise exist. The high jumper could walk under the bar; they choose to jump. The swimmer could get out of the pool; they choose to turn at the wall. The value of sport is inseparable from the voluntary acceptance of constraints that make the embodied achievement meaningful.
What is the relationship between this account of embodied excellence and AI-assisted sport? At first glance, AI seems to pose no threat to embodied excellence: the athlete still runs, jumps, throws, and competes; the AI merely helps them prepare and recover more effectively. But this reassurance becomes less convincing upon examination.
Consider the use of machine learning in talent identification and athletic development. Contemporary AI systems can process biometric, psychological, and physiological data from thousands of young athletes to predict, with considerable accuracy, which individuals are most likely to achieve elite performance. These predictions shape which children receive elite coaching, development funding, and competitive opportunities. The system works by identifying patterns in historical data about elite athletes — patterns that reflect the particular physical, psychological, and socioeconomic characteristics of those who have historically succeeded in a given sport.
This process raises profound questions about embodied agency. The athlete's development is no longer primarily a story of their own striving, error, and discovery — it is increasingly the execution of an algorithmically optimised plan. The phenomenological richness of athletic development — the athlete's own embodied knowledge of their capacities, their creative experimentation, their cultivation of a distinctive style — is at risk of being replaced by data-driven compliance with optimised training regimens. As the philosopher Drew Hyland argued, sport at its best is a mode of responsive openness to the possibilities of human existence. AI optimisation tends toward closure, not openness.
There is also a deeper worry about what we might call the problem of embodied opacity. The most important aspects of athletic excellence — the feel of a perfectly timed tackle, the phenomenological experience of "flow" in competitive performance, the embodied knowledge that a particular movement is right — may be fundamentally resistant to algorithmic capture. If sport's governing bodies come to define excellence in terms that are fully legible to AI systems, they risk systematically undervaluing the aspects of sport that are most distinctively human.
4. Public Trust and the Transparency Imperative
Sport requires public trust to function as a social institution. Spectators must believe that competition outcomes are determined by the relevant athletic capacities, not by corruption, bias, or luck. Athletes must believe that they are evaluated on relevant criteria. Governing bodies must demonstrate that they exercise authority in the service of sport's constitutive values rather than commercial or political interests.
AI systems pose distinctive challenges to each dimension of trust. For spectators, the problem is comprehensibility: if VAR decisions are based on pixel-level geometry calculated by an opaque algorithm, fans cannot evaluate whether the decision was correct or fair. The legitimacy of officiating in democratic societies depends, at least in part, on the possibility of informed criticism by stakeholders who understand the decision-making process. Opaque AI officiating forecloses this possibility.
For athletes, the problem is accountability: if a performance analytics AI recommends that a player be dropped, traded, or retrained, who is responsible for that recommendation? The coach who followed it? The algorithm? The engineers who designed it? The club that deployed it? In traditional human decision-making, responsibility is relatively clear. AI-mediated decisions diffuse responsibility across multiple actors in ways that can make accountability difficult to assign.
For governing bodies, the problem is regulatory capture: AI systems in sport are largely developed by technology companies whose commercial interests may not align with sport's constitutive values. If FIFA uses a goal-line technology developed by a private company, and that company's contract prevents public disclosure of the system's technical specifications, then the governance of an important aspect of sport has been effectively privatised.
5. Distributive Justice and the Access Problem
Perhaps the most profound ethical challenge posed by AI in sport is distributive: who has access to AI performance enhancement technologies, and what are the consequences of unequal access?
The contemporary landscape of AI-in-sport is deeply unequal. The most sophisticated performance analytics platforms, injury prediction systems, and training optimisation tools are available only to the wealthiest clubs and national federations. Manchester City, with its partnership with the data analytics company Second Spectrum, operates in a different technological world from a club in the English Championship, let alone a team in the Zambian Premier League. The United States Olympic and Paralympic Committee's partnership with machine learning companies gives American athletes access to training optimisation tools unavailable to athletes from lower-income countries.
This inequality has direct implications for competitive outcomes. If AI training systems can meaningfully improve athletic performance, then access to these systems constitutes a form of competitive advantage. This advantage is distributed not on the basis of athletic talent or effort but on the basis of economic resources. From a Rawlsian perspective, this is particularly troubling: the least advantaged athletes and nations — those who most need developmental support — are precisely those denied access to the most powerful developmental tools.
The access problem is exacerbated by the data requirements of AI systems. The most powerful machine learning systems require vast quantities of high-quality training data. This data is concentrated in wealthy sports organisations with the infrastructure to collect, store, and process it. Poorer organisations not only lack access to AI tools but are systematically excluded from the data ecosystems on which those tools depend.
6. Toward a Framework for Responsible AI in Sport
Drawing on the analysis developed in the preceding sections, we propose five normative principles for a responsible AI-in-sport framework.
Transparency: AI systems used in officiating, athlete evaluation, or competition governance must be subject to meaningful transparency requirements. Stakeholders — athletes, clubs, fans, and governing bodies — must have access to sufficient information about how AI systems work to evaluate their outputs and hold accountable those who deploy them. This does not require open-sourcing every algorithm; it does require clear articulation of decision criteria, regular independent audits, and mechanisms for stakeholder review.
Contestability: Decisions made by or substantially influenced by AI systems must be subject to meaningful challenge. Athletes and clubs whose interests are affected by AI recommendations or judgments must have access to processes through which they can contest those judgments before appropriately qualified human authorities. The principle of contestability is especially important in high-stakes contexts such as officiating decisions that affect match outcomes.
Embodiment-sensitivity: AI systems designed to optimise athletic performance must be evaluated not only for their technical effectiveness but for their effects on the phenomenological and developmental dimensions of sport. Systems that reduce athletic development to data-compliant execution of optimised plans, or that systematically undervalue forms of embodied excellence that are difficult to quantify, should be considered failures even if they produce measurable performance improvements.
Distributive access: The benefits of AI performance enhancement technologies should not be reserved for wealthy clubs and wealthy nations. Sport's governing bodies have an obligation to ensure that AI tools that can meaningfully improve athletic development are available, on equitable terms, to organisations and athletes who cannot afford commercial market prices. This may require public or cooperative ownership of core AI infrastructure, subsidised access programmes, or mandatory open licensing of AI tools developed with public funding.
Democratic governance: Decisions about which AI systems to deploy in sport, on what terms, and subject to what constraints must be made through processes that give genuine voice to all major stakeholders — athletes, coaches, fans, officials, and communities — rather than being determined unilaterally by technology companies, wealthy clubs, or unaccountable bureaucracies. The governance of AI in sport is ultimately a political question about what sport is for and what values it should embody.
7. Conclusion
Artificial intelligence is transforming sport with a speed that outpaces philosophical and ethical reflection. This paper has argued that the ethical challenges of AI in sport cannot be adequately addressed by technical means alone. They require a philosophical framework that takes seriously the constitutive values of sport — fair play, embodied excellence, and public trust — and evaluates AI systems not merely by their technical performance but by their effects on these values.
The five principles we have proposed — transparency, contestability, embodiment-sensitivity, distributive access, and democratic governance — are demanding. They require governing bodies, technology developers, and sports organisations to accept constraints on their autonomy and to engage in genuine democratic deliberation about the role of AI in sport. This will not always be easy or popular. But the alternative — allowing AI to reshape sport according to the imperatives of commercial optimisation and competitive advantage, without democratic deliberation about the values at stake — risks transforming sport from a human institution that reflects and cultivates human capacities into an algorithmic process in which human participation is an increasingly marginal variable.
Sport at its best is a mode of human self-expression, a celebration of the possibilities of embodied existence, and a theatre of democratic aspiration. Artificial intelligence can serve these purposes — but only if societies make deliberate choices about how to govern it.
References
- Anderson, C. and Sally, D. (2013) The Numbers Game: Why Everything You Know About Football Is Wrong. New York: Penguin.
- Dreyfus, H. (1972) What Computers Can't Do. New York: Harper & Row.
- Floridi, L. et al. (2018) 'An ethical framework for a good AI society', Minds and Machines, 28(4): 689–707.
- Hyland, D. (1990) Philosophy of Sport. New York: Paragon House.
- Loland, S. (2002) Fair Play in Sport: A Moral Norm System. London: Routledge.
- Morgan, W. (2006) Ethics in Sport. 2nd ed. Champaign: Human Kinetics.
- Rawls, J. (1971) A Theory of Justice. Cambridge: Harvard University Press.
- Suits, B. (1978) The Grasshopper: Games, Life and Utopia. Toronto: University of Toronto Press.
- Zuboff, S. (2019) The Age of Surveillance Capitalism. New York: PublicAffairs.