Game Changer: Harnessing Artificial Intelligence in Sport for Development

Abstract

Sport for Development (SFD) leverages sports as a tool to support broader sustainable development goals, particularly in underserved communities worldwide. As Artificial Intelligence (AI) technology advances, its application in SFD offers both promising opportunities and significant challenges in areas such as curriculum design, evaluation, and participant engagement. Through a qualitative survey of experts and practitioners analysed through Thematic Analysis (TA), this paper explores perspectives on the potential of AI to enhance the delivery and management of SFD initiatives, as well as potential risks and needs in the field. Key perceived benefits include compensating for deficient organisational capacities and supporting the performance of both administrative and conceptual tasks. Potential risks include the propagation of increasingly generic approaches to SFD programming, loss of critical thinking skills, and concerns around participant safeguarding. To mediate this, exchange, education, and SFD-specific policies are seen as crucial.

1. Introduction

Since the launch of ChatGPT in November 2022, the usage of Artificial Intelligence (AI), otherwise known as Generative AI (GenAI) tools, has grown substantially. In particular, there has been extensive discussion about the potential role AI could play in achieving various Sustainable Development Goals (SDGs) and the overall sustainability of AI itself (van Wynsberghe 2021; Vinuesa et al. 2020). Many fields dedicated to supporting these goals, such as social work, education (Farrokhnia et al. 2024), and health promotion (Ha et al. 2024), have likewise further engaged with the implications of these tools within their spheres of action.

Sport for Development (SFD) is another such area that is gaining growing attention for the potential of AI. Broadly defined as the intentional use of sport or physical activity to achieve sustainable development goals, a growing number of applied resources and consultants have begun offering advice concerning the usage of AI in SFD (SportandDev.org 2024; Sheety 2024). Within SFD, programmes often deploy sport as a means to support developmental goals, either by using sport-specific pedagogical approaches or by using sport as an attractive hook to direct participants towards other forms of programming, such as workshops, volunteering, or counselling (Moustakas 2024). Structurally speaking, the majority of organisations delivering SFD activities are not-for-profit, civil society organisations that often are confronted with limited resources and irregular funding, thus mitigating their ability to tend to broader organisational or administrative concerns, such as reporting, communications, and monitoring (Whitley et al. 2020; Svensson and Hardie 2024).

With this in mind, proponents of AI in the SFD sector suggest that AI-based tools can remedy some of these deficits by supporting tasks such as curriculum development, fundraising, education, and participant engagement (SportandDev.org 2024; Sheety 2024). Research in neighbouring fields, such as sport management, coaching, and education, likewise highlight these potential applications, while also cautioning against the many risks associated with AI-based tools (Sperlich et al. 2023; Farrokhnia et al. 2024; Ha et al. 2024; Glebova et al. 2024).

Despite the apparent promise of AI for SFD, AI-based tools have been evolving at a rapid pace and present potential threats. As a result, numerous authors in sport and other fields have suggested that continuous research within specific sub-domains or contexts is needed to provide fine-grained, transferable insights (Sperlich et al. 2023; Glebova et al. 2024). The need for more contextualised research on the role of AI is especially pronounced in a field such as SFD, as programmes often combine activities and perspectives from a wide range of disciplines, including social work, pedagogy, and sport sciences (Whitley et al. 2022; Bauer 2024). For instance, sport-based activities often rely on pedagogical concepts concerning experiential learning, while the programmes themselves are often delivered by individuals with backgrounds in community development or social work (Moustakas 2024; Dušková et al. 2024). Furthermore, SFD is often characterised by inherent power imbalances—such as among marginalised communities, funders, and programmes implementers—adding a further layer of complexity to the usage of any new technological tool. Extensive critical work has already addressed how the knowledge, skills, and needs of marginalised communities are sidelined within SFD (Harris and Adams 2016; Nicholls et al. 2011; Darnell and Hayhurst 2011), and it is essential to consider how the potentially massive technological shift caused by AI may further influence this disparity. In short, as Svensson and McSweeney (2022) note, “the importance of technology to SFD organisations, practitioners, and researchers is thus an area requiring significant attention in the SFD field” (p. 159).

Against this background, using the results of a qualitative expert survey, this paper explores the potential opportunities, risks, and needs associated with AI within the SFD context. Moving forward, this paper progresses in four steps. First, I will review the general literature on technology within SFD and connect it to current debates on AI usage. Second, I will present the design and methodology of this study. Third, I will outline the main results emanating from the expert survey. And, finally, I will critically reflect on these findings in order to develop recommendations for future practice and research.

2. Technology, AI, and Sport for Development

There is a nascent but growing body of literature concerning the role of digital technologies within SFD, as many authors recognise the potential of these technologies to support the development outcomes targeted by SFD programmes while also compensating for the sometimes deficient material resources of these programmes (Svensson and McSweeney 2022). Broadly speaking, extant literature at the intersection of SFD and technology has mainly concerned itself with two broad areas: organisational communications and monitoring and evaluation. In terms of communications, various studies have looked at how web or digital communications can support participant engagement, funder outreach, and overall programme dissemination (Svensson et al. 2015). For instance, various studies highlight how the use of social or mobile media channels can assist with participant outreach, especially in remote or poorly connected areas (Svensson and McSweeney 2022). Technology can also be used to reach potential supporters or donors, as in the case of SurfAid or Skateistan, who deploy short videos or other media to create empathy and, ultimately, financial support for their work (Thorpe and Rinehart 2013).

As for Monitoring and Evaluation (M&E), the existing literature highlights how the use of digital tools, such as Salesforce or other software, are being used to track programme data and feedback more systematically. For example, within Grassroots Soccer’s work, such software was used to track attendance, demographic, and HIV testing data within a South African project, eventually contributing to a related academic publication (Hershow et al. 2015). Another emerging area in which technology is seen to contribute to M&E relates to its use as a vehicle for participant-led data collection. In particular, researchers note how tools such as photovoice or digital storytelling can help empower participants during data collection while also eliciting rich information (McSweeney 2023).

What is still missing from the extant SFD literature so far is specific consideration of the role AI technology can play in these areas or others. As highlighted above, SFD is situated at the intersection of various disciplines and often targets marginalised populations, thus creating a fairly unique mix of approaches and challenges. Though some applied work suggests that AI can be used in SFD to support monitoring, data analysis, curriculum design, or administrative tasks (SportandDev.org 2024; Sheety 2024), more academic work has yet to look at this specific field. Nonetheless, other authors from connected fields, including social work, education, and sport management, have reflected on the opportunities and risks associated with AI in their domains. And, as these areas all share overlaps with SFD, they are worthy of brief discussion here. Overall, this body of literature highlights the potential for AI to support organisations by increasing access to in-depth information, streamlining administrative tasks, supporting with data analysis, and assisting the development of pedagogical materials (Farrokhnia et al. 2024; Ha et al. 2024; Glebova et al. 2024; Sperlich et al. 2023). A particular salient use case is how AI can potentially enhance educators’ or trainers’ pedagogical abilities (Chiu et al. 2023; Ha et al. 2024). For instance, as Farrokhnia et al. (2024) noted, AI-based tools can support educators who have less teaching experience and assist in the design of activities, quizzes, and other examinations. Similarly, others have documented how AI is already being used to develop physical education courses, athletic training regimens, and performance analyses (Glebova et al. 2024; Ha et al. 2024). However, concerns persist about how the usage of AI in such educational contexts may fail to deliver context- or student-specific education, creating disengagement and lack of connection amongst learners (Chiu et al. 2023). In addition, there are growing concerns around how the continued growth and uptake of AI-based tools may simply reinforce existing inequalities based on the ability of individual educators or students to access, afford, and use these technologies (Chiu et al. 2023).

In a similar vein, these authors caution that over-reliance on AI-based tools carries a significant risk of decreased critical (or higher-order) thinking skills, as individuals may become complacent given the ease of use associated with these tools (Sperlich et al. 2023; Farrokhnia et al. 2024). In particular, some authors have highlighted how AI tools often generate generic information or analyses that lack a fine-grained understanding of the local context, and that higher-order thinking skills are needed to re-contextualise AI-generated content (Farrokhnia et al. 2024; Ha et al. 2024). Furthermore, there is an awareness that AI tools, as they are based on human-generated knowledge, often risk reproducing existing biases related to ethnicity, gender, or other social factors (Farrokhnia et al. 2024; Ha et al. 2024). More philosophically oriented critiques also argue that these kinds of risks are exacerbated by the mysticism surrounding AI, whereby numerous prominent figures posit that AI will match and surpass human intelligence, thereby fostering a misplaced trust by suggesting that these tools are, or will soon be, superior (Böhm and Sammet 2024).

The above literature and analyses, though useful, come from a variety of areas that, while close to SFD, do not encompass the interdisciplinary and multisectoral nature of SFD. Thus, in the following, we seek to extend our understanding of the potential opportunities and risks of AI by specifically engaging with experts from the SFD sector.

3. Methodology

3.1. Instrument

An online, qualitative expert survey was designed and deployed to assess perspectives of the potential opportunities and risks associated with AI usage in the SFD context. The choice was made to use an online qualitative survey, as it allowed me to reach out to a diverse range of experts while allowing them to respond flexibly and without pressure, minimising many of the logistical challenges associated with scheduling, videoconferencing, and internet connectivity associated with the broad range of international experts targeted (Braun et al. 2021).

The survey was designed based on the guidance provided by Braun et al. (2021) and Thomas et al. (2024), concerning the usage of online surveys as qualitative research tools. In line with the research questions, a first survey was designed and sent to five individuals for piloting. These individuals not only provided their responses to the survey but also offered separate feedback on the clarity, completeness, and quality of the questions, as well as on any potential technical issues. After the pilot round, a final survey was put together. This survey included two main parts. The first part focused on collecting general background information about the respondents, including the type of organisation they work for, their geographic location, and their overall experience with AI tools. The second part, which is summarised in

Table 1. Overview of main open-ended questions.

Questions
What benefits or opportunities do you think SFD organisations can obtain from using AI-based tools? What risks or threats do SFD organisations face from using AI-based tools? What skills, resources, or support do SFD organisations need to maximise the benefits of AI-based tools? How can SFD organisations minimise the risks or threats associated with AI-Based tools?

, focused on four main, open-ended questions related to opportunities, risks, and organisational needs concerning AI within the SFD context. At the end, respondents were also given space to integrate any further comments or thoughts not captured by these questions. Throughout the survey, guidance was likewise provided to establish a clear, shared understanding of AI, and prompt respondents to provide detailed answers.

3.2. Data Collection

A combination of purposive and snowball sampling was used to target experts and practitioners involved in SFD. The individuals targeted were defined under three broad categories. The first group targeted featured academics, consultants, and members of international SFD organisations who could provide a bird’s eye view of the field and who play a key role in setting the agenda for the field. The second group of experts included individuals with a specific expertise at the intersection of SFD and technology, thus allowing the survey to capture the voice of those most engaged with topics like AI, digitalisation, and innovation. Finally, a third group of practitioners from organisations implementing SFD activities were targeted. These individuals were seen as critical, as they will be on the front lines of using AI within the day-to-day work of SFD.

With these criteria in mind, an initial list of approximately 99 experts was developed. This list was developed through a scan of major international sport-for-development organisations, thematic online networks, as well as my own professional contacts. All individuals from the initial list were contacted directly via e-mail or social media. In addition, the survey was further distributed via a number of thematic SFD networks, thus allowing me to capture a broader array practitioners and implementers (e.g., SportandDev.org, Ashoka, CommonGoal). The survey ran from 18 November 2024 to 8 December 2024.

In total, 43 responses were collected from individuals in 28 countries, including Germany (9) and the United States (3), as well as Australia, Canada, Czech Republic, India, North Macedonia, and South Africa (2 each). Of these responses, 13 individuals indicated working for an international SFD organisation, 7 indicated working for a regional/national SFD organisation, while another 7 indicated working for an academic or research institution. Other organisational categories represented included consultancies or thinktanks, international sport federations, national sport federations, and governmental bodies.

On average, the survey took approximately 20 min, and open-ended responses were 207 words in length. All participants were informed about the nature of the survey and how their data would be used, and they provided informed consent directly within the survey.

3.3. Data Analysis

Thematic analysis was used to guide the coding and analysis of the data (Braun and Clarke 2022). This approach was chosen, as it philosophically aligns with the approach taken in open-ended surveys and is likewise suitable for use in both smaller and larger datasets. Thematic analysis focuses on developing themes—coherent patterns “of shared meaning organised around a central concept” (Braun and Clarke 2022, p. 77)—from the data. Developing such themes moves the research beyond mere summarisation or description and enables the analysis to highlight the ways in which aspects of a phenomenon may be connected (Braun and Clarke 2022, 2006; Nowell et al. 2017).

My analysis employed the six steps defined by Braun and Clarke (Braun and Clarke 2006, 2022), namely familiarisation, generating initial codes, searching for themes, reviewing themes, defining themes, and writing the report. Although this data analysis is textually presented in a linear fashion, it was, in fact, an iterative process that involved constant movement back and forth between phases.

After closing the survey, I read all of the answers, scanning for descriptive participant information, made initial notes about the data, and began developing first ideas for potential codes. Following familiarisation, I re-read the responses and engaged in a first round of coding. At this stage, I used MaxQDA 2024 to manage the dataset, collate notes and memos, assign codes, and perform further analysis. Here, codes should be understood as descriptive labels assigned to segments of the qualitative data that allowed me to sort and reference the data systematically. I generated codes inductively based on the data, and these codes were defined (and refined) throughout the process. Specifically, I developed codes related to the opportunities, risks, and potential needs associated with AI usage within SFD organisations. Following this initial round, I further refined and revised the coding by reviewing the code descriptions and, if needed, merging codes to avoid duplication or exceedingly small distinctions. In addition, I re-read the data extracts and used built-in search functions within MaxQDA to ensure that the entire dataset was appropriately coded.

Once the coding was complete, I moved on to analysis and theme development. This involved reviewing the data, coded segments, notes, and reflections as a whole, with a view towards identifying patterns of meaning within the data. To support this process, I explored the interrelations among the codes using built-in visual tools within MAXQDA, including word clouds, code relations, and code maps (see, e.g., Gizzi and Rädiker 2021), and began drafting theme summaries within the software as well. This process allowed me to go from summarising the data to developing themes that illustrate the shared patterns of meaning within the dataset (Braun and Clarke 2022). I iteratively reviewed these themes, constantly comparing them to the coded data and visual summaries, and I further discussed my findings with colleagues to ensure consistency and foster further reflection. From this iterative process of analysis and reflection, I settled on three main themes related to the core topic of this paper, which I present in the upcoming section.

4. Results

Overall, the three themes developed—“Keeping Things in Context,” “Bridging or Widening the Gap”, and “Education and Internal Control”—show the inherent benefits and tensions, as well as the needs of SFD actors, as they relate to AI. These three themes are presented purely as results and will be further contextualised against the existing literature in the subsequent discussion section.

4.1. Keeping Things in Context

One key benefit of AI tools lies in their perceived ability to organise, summarise, generate, and repurpose information or data at scale. For many respondents, there was a clear focus on potential applications, such as managing monitoring and evaluation data, generating curriculum ideas, and supporting (multilingual) communication efforts. Monitoring and evaluation were especially prominent as an area of potential use, with respondents suggesting that AI could support the delivery of “automated and personalised surveys” (Respondent 3), assist with “data analysis” (Respondent 41), and even deliver “predictive analytics to anticipate programme outcomes” (Respondent 9).

Recognising the precarious and intense competition for funding in the field, participants also noted how AI tools could assist in fundraising by supporting grant writing, especially for non-native English speakers, and supporting the overall design of grant proposals:

I believe the opportunity for organizations and practitioners, particularly those who do not speak English as their first language, would benefit in regard to using AI for grant-writing, which is dominated by Western funders/donors who sometimes only accept proposals in English. I believe using AI may support organizations in effectively explaining their programmes

(Respondent 27).

Yet, the delegation of these information- and content-generating tasks to AI tools was likewise seen to carry inherent risks, especially in terms of homogenisation and the deterioration of critical thinking skills. Regarding the first point, there was consistent recognition that SFD programming is highly context-sensitive and engages with marginalised groups. This means not only ensuring privacy and safety standards but also handling the oft-generic responses generated by AI tools with care and sensitivity to ensure that they are appropriate and useful. Furthermore, many respondents noted that AI-generated content may lack the nuance and local relevance required for effective SFD programming:

Most importantly, it shouldn’t lose sight of the human factor and importance in driving the mission of SFD organisations (developing communities and individuals through sports). In this sector, the development comes through the relationships with coaches, through the sense of community that is being built, through a safe physical and emotional environment. Those aspects remain, in my opinion, hard to reach with AI

(Respondent 3).

As for the second risk, participants highlighted how an overreliance on AI tools may eventually dull higher-order thinking skills. As more complex tasks become delegated to AI-based tools, the concern is that individuals will engage their critical thinking and creativity less, leading to a multiplication of unreflected, decontextualised approaches that “underestimate the role of the personal touch” in SFD (Respondent 8). Following this, some argued that this could even create a situation where content and approaches in the field become increasingly homogenised or generic, “creating a situation in which organisations fall into a herd mentality instead of making social aims relevant to their region or context” (Respondent 24).

4.2. Bridging or Widening the Gap?

The respondents did not merely consider benefits or risks in terms of individual organisational tasks but also for the field as a whole. A clear tension in the dataset revolved around the potential of AI, on one hand, to compensate for the often-limited capacity of SFD organisations while, on the other hand, exacerbating existing inequalities in the field.

Given the current relatively low-cost, easy to access nature of many tools, there was a sense that AI tools could provide a low-threshold solution for under-resourced organisations. As highlighted above, and recognised by respondents, SFD is often delivered by small organisations with limited capacities, leading to a perceived opportunity for “reduced costs” where “almost every stage of the project cycle can be improved when struggling with low resources/capacity” (Respondent 23). Beyond improving the execution of individual tasks, there was also a perception that these tools could help streamline many time-costly administrative tasks that tend to distract from fundraising, programme conceptualisation, and delivery:

A lot of SDP organization[s] are run by a handful of people, so tasks that perhaps at one point seemed impossible to find time for, one can now find time even if the resources aren’t there, especially in terms of streamlining

(Respondent 12).

[AI] will enable more organisations to access otherwise expensive and time-consuming sophisticated tools and insights. As such, we can expect AI to level the playing field for smaller organisations or those with limited resources, ensuring that they operate on par with larger entities

(Respondent 9).

In contrast with the optimistic predictions above, there was a recognition that the usage of these tools is dependent on having the “necessary skills or infrastructure to benefit from AI tools” (Respondent 30), including sufficient financial resources, reliable internet connectivity, and adequate technological skill. The dependence on such resources could mean that already well-equipped organisations will be the ones to benefit from AI, while others will merely be left behind, “Increasing the digital divide between more established and those less established (or with more/less resources, skills, knowledge, and access to tools). Those who use AI may end up being miles ahead” (Respondent 43). For instance, the emergence of AI-driven forms of fundraising, reporting, and administration could end up making “some knowledge and expertise (…) obsolete” (Respondent 3). Ultimately, as one respondent concisely outlined, managing an equitable allocation of resources is crucial to ensure that (new) divides are not created: “if the access to AI is given equally to SFD organisation from LMICs like in Global North, it could increase the accessibility of organisations and help in compete for various resources with other organisations across the globe” (Respondent 26).

4.3. Education and Internal Control

As many respondents recognised, managing both organisational and sectoral risks will require a combination of efforts and solutions. By far, however, the most prominent solution concerns ensuring comprehensive training and education programmes to build knowledge and skills among staff and stakeholders. In other words, there is a “need to empower people to learn about how it is going to best serve their organisation” (Respondent 1). This includes training on the fundamentals of AI, its potential benefits, and limitations, as well as ethical considerations around data privacy and responsible use. Respondents further emphasised the need for “bespoke”, context-specific training tailored to the needs of SFD and focus on how AI tools “can be used specifically in the SFD context” (Respondent 15), rather than generic, one-size-fits-all programmes. The respondents also expressed the desire that this education should be complemented by widely available educational resources, such as materials “available online” (Respondent 12) or guides on ethical usage, as well as by constant exchange and reflection amongst actors in the field. Indeed, exchange is an important component here, as numerous respondents highlighted the need for cross-sectoral collaboration, or as one individual put it, that “reflective practice needs to be intentional” (Respondent 5).

Complementing this, the respondents recognised the need to develop both personal behaviours and internal policies to ensure a shared understanding of AI’s usefulness and limitations. Appropriate individual usage and a strong internal policy framework were seen as crucial to limit and navigate the many pitfalls associated with AI, including those regarding privacy, safety, environmental impact, and information reliability. Partially, this means that individuals should adopt critical practices whereby individuals constantly “check sources and evidence” while combining “human and AI skills to improve the work” (Respondent 36). Parallelly, there was also recognition that organisations themselves should underpin and guide those individual behaviours by conducting readiness assessments, establishing policies, and creating roadmaps to support the responsible internal use of AI: “Implement them in a considered and responsible way, with a thorough preparatory review as to how exactly they will be used within the organisation, when, by whom, etc.” (Respondent 8).

5. Discussion and Conclusions

Through the responses and analysis depicted here, it is evident that actors in the SFD field see many benefits, risks, and needs as they relate to the many emerging AI tools and technologies. Of course, the responses here represent only a subset of the SFD field. The survey captured predominantly English-speaking voices and was mediated by my own contacts and awareness of the field. Future work in different languages, regions, or networks would do well to integrate further voices or perspectives and investigate how specific organisations use and navigate AI in their day-to-day work. Nonetheless, some clear results can be extracted from the existing dataset. What is notable is that most of the points highlighted, which are summarised in

Table 2. Summary of benefits, risks, and solutions captured in the data.

Benefits	Streamlining and automation of management tasks Compensating for lacking internal resources Supporting monitoring, data collection, and evaluation Designing communication and marketing materials Supporting grant and proposal writing Supporting programme and curriculum design Translation and cross-cultural communication
Risks and Threats	Reinforcing inequalities between regions and organisations Lack of resources and skills Dulling creativity and critical thinking Potential threats to participant privacy and safeguarding Environmental impact
Solutions and needs	Sector-specific education and training Opportunities for exchange and reflection Tools, templates, and learning materials Internal assessments and policies

, reflect the core structure and requirements of the field. In other words, the potential benefits and risks are contextualised within the existing resources, structures, and relationships underpinning SFD, and the respondents actively consider how AI might affect these existing dynamics, especially as it concerns existing resource gaps within the field.

Several respondents emphasised the potential benefits regarding data collection, monitoring, evaluation, and grant writing, all of which tend to be strongly interrelated within SFD. So-called evidence of positive outcomes is highly prized within SFD, as it serves to bolster organisational reputations and increases the odds of receiving future funding (Harris and Adams 2016; Webb and Richelieu 2016). In contrast, there are also significant concerns about how potential overreliance on AI will reduce the context-specific, authentic programming needed to generate successful outcomes. Despite the proliferation of frameworks and theories of change within the field, even these attempts at some level of generalisation highlight the central role of context-specific, personalised programming and approaches in the success of any programme (Coalter 2013; Verkooijen et al. 2020). Though studies concerning the impact of AI on critical thinking are mixed, it is clear that careful consideration and management are needed (e.g., Darwin et al. 2024). Furthermore, as SFD often features a strong pedagogical component, it is important to carefully consider how AI-generated content or curricula affect the socio-emotional connection of participants (Chiu et al. 2023). Human relationships and personal connection are often presented as crucial within SFD—with the coach or social worker playing a central role—and AI may pose unique challenges in that respect (cf. Jeanes et al. 2019).

Taking a field-wide perspective, there are fears that a new digital divide will open up, further disadvantaging under-resourced organisations—many of whom are in the Global South. As the use of AI tools is increasingly resource-intensive, including in terms of internet connectivity, skill, and cost, organisations with fewer resources may opt to not invest in this area despite its potential benefits. This concern very much aligns with previous critiques that knowledge, funding, and implementation are primarily driven by well-funded Global North organisations that do not fully consider the needs, expertise, or knowledge at the local level (Harris and Adams 2016; Nicholls et al. 2011; Darnell and Hayhurst 2011). More broadly, this may also reflect a continuation of previously identified trends, where the Global South has largely caught up with the North in terms of basic capacities (e.g., primary education, cell phone coverage) but faces a widening gap in terms of more advanced capacities (e.g., post-secondary education, high-speed mobile internet, etc.) (UNDP 2019). Future work should consider how differing types of SFD organisations navigate and deploy these AI tools while being mindful of these potentially important resource differences.

The solutions proposed to address these risks primarily rely on individual and organisational actions, including adopting new behaviours, exchange, education, and internal policies. These recommendations largely reflect those in the other SFD-adjacent literature (Sperlich et al. 2023; Glebova et al. 2024). What is most notable here, however, is the lack of any explicit mention of support, policy, or regulation from technology firms or governments. Though, admittedly, the survey questions did not explicitly ask for the SFD sector’s external needs, the wording did not exclude them either. This may suggest that the field views itself as a somewhat powerless, passive entity that is on the receiving end of these new developments but has no expectation to be able to shape them. Yet, as previous work has highlighted, there may be significant benefits for SFD organisations to advocate more strongly for certain policies or changes, especially since those policies can have a meaningful impact on both the organisations themselves and their programme participants (Sanders 2016; Moustakas et al. 2025). Forming coalitions, engaging with policymakers, and raising public awareness are some potential avenues for SFD organisations to gain more agency against the rising tide of AI developments. Connected to this, and given the growing importance and reach of AI tools, future research and practice could consider how the field may want to position itself in the face of these recent technologies. After all, as the data here underscore, there are a real individual and ethical risks around programming quality, privacy, participant safeguarding, and environmental impact, and the field may benefit from developing a coherent, united strategy to address these.

Looking ahead, it is worth contemplating how these AI tools will develop and how the SFD field can respond to these developments. Many popular tools and software, including social media platforms, dating apps, and more, have undergone a progressive process of "entshittification" (i.e., platform decay), whereby they initially offer tremendous values to users until they eventually impose new constraints and processes that reallocate most of that value to the operating companies (Doctorow 2022). In turn, this has created a slew of ad-filled, algorithm-driven tools that not only offer a declining user experience, but also actively contribute to division, create echo chambers, and sometimes enable political violence. Yet, at the same time, escaping these platforms has proven difficult due to network effects and switching costs (Doctorow 2022). This potential risk underscores the importance for SFD organisations, as well as organisations in general, to have clear policies and standards that outline not only how they will use these emerging tools but also under what quality and ethical circumstances.

In conclusion, this study highlights the potential of AI within sport for development while underscoring critical challenges that demand attention within the field. By streamlining administrative tasks, supporting curriculum design, and facilitating multilingual communication, AI is seen as providing significant opportunities for under-resourced SFD organisations. However, these benefits are counterbalanced by perceived risks around the reinforcement of existing inequalities, ethical concerns regarding data privacy, and the potential erosion of critical thinking skills. Addressing these challenges requires targeted education and training, robust internal policies, and advocacy for equitable access to AI tools. As the field evolves, future research must explore the longer-term impacts of AI on SFD practices and participants, ensuring that technological advances align with the sector’s overarching developmental goals.