The Impact of AI in Sustainable Development Goal Implementation: A Delphi Study

Abstract

Artificial intelligence emerges as a powerful catalyst poised to reshape the global sustainability landscape by facilitating the attainment of Sustainable Development Goals (SDGs). This comprehensive Delphi study meticulously probes the insights of domain experts, shedding light on the strategic utilization of AI to advance these critical sustainability objectives. Employing rigorous statistical techniques, encompassing measures of central tendency and interquartile deviation, this research scrutinizes consensus dynamics among experts and elucidates potential variations in their viewpoints. The findings resoundingly convey experts’ collective positive perspective regarding AI’s pivotal role in propelling the SDGs forward. Through two iterative rounds of extensive discussions, a compelling consensus crystallizes—AI indeed exerts an overall positive impact, exemplified by a robust mean score of 78.8%. Intriguingly, distinct SDGs manifest varied propensities toward AI intervention, with Goals 6, 7, 8, 9, 11, 13, 14, and 15 basking in the radiance of highly positive impacts. Goals 1, 2, 3, 4, 5, 10, and 12 exhibit positive impact scores, indicating a juncture ripe for positive advancements. Meanwhile, Goal 16 and Goal 17 languish with neutral scores, signifying a juncture demanding nuanced deliberations about AI’s impact on peace, justice, and strong institutions as well as on partnerships for the goals, respectively. This paper underscores AI as a formidable instrument poised to address humanity’s most pressing challenges while harmonizing seamlessly with the overarching SDG objectives. It gracefully dovetails into established practices across pivotal domains such as health, education, and resilient infrastructures, amplifying the collective global endeavor to navigate the path toward a more sustainable future.

1. Introduction

This study begins by highlighting the Sustainable Development Goals (SDGs) as a global imperative, then examines AI as a key driver for achieving these goals. It explores AI’s advancements and its dual impact on SDG progress, concluding with the importance of ethical AI practices and international collaboration for sustainable development.

The United Nations’ Sustainable Development Goals delineate a global imperative to confront and mitigate the most urgent social, economic, and environmental dilemmas that threaten our world’s equilibrium by the year 2030 [1]. These goals are not just aspirations; they are a blueprint for collective action, demanding innovative approaches and transformative technologies to drive progress. This era is witnessing the convergence of such transformative technology—AI—with the SDGs, heralding an unparalleled prospect to tackle intricate barriers that obstruct the path toward realizing sustainable development [2]. Amidst this ambitious global agenda, artificial intelligence, typified by its prowess in mimicking human cognitive functions, has undergone a significant renaissance in recent epochs. This resurgence, propelled by seminal innovations in machine learning, natural language processing, computer vision, and robotics, has catalyzed transformative shifts across diverse sectors and redefined societal paradigms [3]. This renaissance in AI technology positions it as a potent ally in achieving the SDGs. Standing at the intersection of cutting-edge technology and global sustainability, the ability of AI to sift through colossal data sets, discern intricate patterns, and extract actionable insights has been demonstrated as a pivotal mechanism in unlocking novel avenues for sustainable growth [4]. By amplifying human faculties, AI equips policymakers with empirical insights to craft targeted policies and interventions aligned with specific SDG objectives. Moreover, AI-infused analytics furnishes the means to gauge progress with precision, pinpoint deficiencies, and fine-tune strategies dynamically, thereby amplifying the efficacy of developmental endeavors [5]. This technological prowess transforms AI into a powerful tool for both monitoring and advancing the SDGs.

Within the SDGs’ framework, AI unfolds a spectrum of diverse solutions encompassing all 17 goals [6]. Ranging from poverty alleviation and healthcare enhancement to climate change mitigation and gender parity, AI methodologies harbor the potential to redress prevailing challenges and instigate pioneering solutions [7]. In essence, AI can be a double-edged sword, capable of both enabling and impeding the realization of SDGs, thereby underscoring the significance of an inclusive and rights-based approach to digital transformation [8]. The United Nations Development Programme (UNDP) has underscored the pivotal importance of evaluating countries’ readiness for the adoption of AI and the establishment of robust data governance frameworks to facilitate the responsible deployment of AI in support of the SDGs [9]. International agreements and frameworks provide invaluable guidance on ensuring the ethical development of AI, particularly within the context of advanced nations [10]. These agreements serve as the bedrock upon which the ethical integration of AI technologies is promoted [11]. Assessing readiness for AI adoption and fostering the responsible development of AI stand as pivotal steps in the effective incorporation of AI into sustainable development endeavors [12]. Ref. [7] provides relevant evidence suggesting that AI may act as an enabler across 134 targets (79%) encompassing all SDGs, primarily through technological advancements that can potentially overcome existing limitations. However, it is important to acknowledge that 59 targets (35%), distributed across all SDGs, may experience adverse consequences stemming from AI development [7]. The SDGs are classified into three overarching categories, corresponding to the pillars of sustainable development, namely Society, Economy, and Environment [13]. After numerous studies on the impact of AI on sustainability, many firms infuse AI into developing their products to improve their SD practices [14]. Nevertheless, it is essential to note that studies supporting AI as an enabler or inhibitor for SDGs were conducted within controlled environments, rendering it challenging to extrapolate these findings to real-world scenarios [7].

Given the ever-evolving societal, environmental, and economic dynamics, maintaining a dynamic feedback loop with AI remains imperative [15]. Collaborative efforts and ongoing impact assessments are critical to maximizing the benefits of AI while aligning with the SDGs. Nations at the forefront of AI adoption demonstrate their potential to foster economic growth and improve livelihoods, thus signaling the importance of actively embracing AI technologies [16]. For the 17 goals, starting with SDG 1, No Poverty, AI holds promise in identifying impoverished areas and tailoring interventions. However, it is a dual-edged sword as concerns about potential job losses due to automation arise. Striking a balance between economic advancement and social well-being becomes imperative [7]. Moving to SDG 2, Zero Hunger, AI’s ability to enhance agricultural productivity through precision farming is valuable. Yet, applying it effectively in resource-limited regions presents notable challenges. To truly address hunger, AI-driven solutions must reach those who need them the most [7,17]. In the realm of SDG 3, Good Health and Well-being, AI’s role in advancing disease diagnoses and treatment holds immense potential. However, it must navigate privacy concerns associated with patient data. Balancing technological advancement with ethical considerations is crucial in this context [7,17]. SDG 4, Quality Education for all, sees promise in personalized learning with AI [18]. But it also raises concerns about exacerbating inequality if access remains uneven. Ensuring equitable access to educational AI tools is pivotal in achieving this goal [7,18]. In the context of SDG 5, Gender Equality, AI’s capacity to identify and rectify biases within systems is valuable. Yet, management is crucial to avoid perpetuating existing societal biases. To promote gender equality effectively, AI must be a tool for empowerment and not discrimination [19]. SDG 6, Clean Water and Sanitation, benefits from AI in predicting water shortages and managing resources. However, challenges related to costs and data limitations may hinder widespread implementation. Addressing these challenges is essential to ensure access to clean water for all [20]. SDG 7, Affordable and Clean Energy, sees AI’s role in enhancing energy efficiency as commendable. However, the energy-intensive nature of large-scale AI systems necessitates attention. Striking a balance between energy efficiency and environmental impact is critical [21,22]. Regarding SDG 8, Decent Work and Economic Growth, AI’s potential to drive economic growth through new industries is noteworthy [23]. Nevertheless, addressing job displacement due to automation is essential. Supporting workforce transitions is vital in this journey toward equitable economic prosperity [16]. In the context of SDG 9, Industry, Innovation, and Infrastructure, AI’s contributions to innovation and productivity are valuable [24]. However, careful consideration is necessary to prevent potential job displacement in certain sectors. Innovation must be coupled with strategies for inclusive growth [7,16,24].

Moving to SDG 10, Reduced Inequalities, AI’s ability to identify patterns of inequality holds significance [25]. Yet, managing its implementation cautiously is imperative to prevent the exacerbation of the digital divide [26]. Bridging the gap between AI haves and have-nots is a key challenge [27]. SDG 11, Sustainable Cities and Communities, benefits from AI’s optimization of urban planning and management [28]. Nonetheless, challenges in resource-poor settings may limit its full potential. Tailoring AI solutions to suit diverse urban contexts is vital in building sustainable communities [29]. While AI enables more efficient resource utilization for SDG 12, Responsible Consumption and Production, concerns exist about its potential contribution to increased consumption and waste [25]. Striking a balance between efficiency gains and responsible consumption is essential [30]. In SDG 13, Climate Action, AI’s capacity to model and predict climate change impacts aligns well. Still, careful consideration must be given to its energy usage. Developing energy-efficient AI solutions is crucial in the fight against climate change [31]. SDG 14, Life Below Water, finds AI’s role in monitoring marine pollution and aiding conservation efforts [25]. However, practical implementation and data collection remain challenging. Overcoming these obstacles is pivotal in protecting our oceans and marine ecosystems [25,32]. Similarly, SDG 15, Life on Land, benefits from AI’s role in monitoring deforestation and supporting conservation efforts [33]. Yet, data limitations may affect its effectiveness. Enhancing data collection and analysis capabilities is essential for preserving terrestrial ecosystems [32,33]. For SDG 16, Peace, Justice, and Strong Institutions, AI’s assistance in conflict prediction and justice delivery holds value [34]. Nevertheless, ensuring its responsible use in surveillance and privacy is essential. Safeguarding human rights and ethical AI deployment must go hand in hand [35]. Finally, in SDG 17, Partnerships for the Goals, AI’s potential to enhance collaboration through data sharing and analyses is valuable [36]. However, addressing unequal access to AI technology is crucial to prevent the exacerbation of inequalities. Building inclusive partnerships is fundamental to achieving shared global objectives [37].

In each SDG, AI presents opportunities and challenges, reminding us of the need for responsible and equitable AI deployment in our journey toward a sustainable and prosperous future [38]. The establishment of a coherent policy framework is pivotal in promoting the adoption of AI to support the SDGs. Such a framework should encompass data governance, capacity-building, and a robust regulatory structure tailored to local norms. This research aims to dissect the symbiotic relationship between AI and the SDGs, exploring both the promises and pitfalls of this transformative partnership. Recommendations are proffered to policymakers, stakeholders, and practitioners to harness AI’s potential in steering sustainable development. By judiciously embracing AI, while adhering to principles of inclusivity, equity, and sustainability, countries can traverse the intricacies of the contemporary epoch and forge a thriving and inclusive future for all.

2. Materials and Methods

A qualitative perspective was employed as the philosophical stand for this study, given its exploratory nature. The Delphi technique was specifically adopted for this study. This approach encompasses the identification of the research problem, followed by the selection of an expert panel and briefing of the experts about the process. According to previous research, the Delphi method is a qualitative methodological approach aimed at achieving consensus among a group of experts on a particular subject [39,40]. For this study, the Delphi technique was applied through two rounds of data collection: the exploratory phase of brainstorming, and the validation and consensus phase of redefining and prioritizing suggested ideas. These were followed by qualitative and quantitative analyses at the end of the first and second rounds. The process culminated in the interpretation of insights and presentation of results, outlining a convergence of opinions from the panel of experts [41]. The adoption of AI for SDG targets and goals is deliberated on, respectively. Once a good consensus has been reached, an overall perspective of the experts is given for the complete potential contribution of AI to the SDGs. The conceptual framework (as shown in Figure 1) shows how this study would be carried out.

The figure above shows the framework for this study. A target (in the diagram) is a specific objective that defines measurable outcomes necessary to achieve the SDGs set by the United Nations for global development by 2030 [13]. The targets in the diagram are represented by “T” with a number indicating the number of targets within a goal. For example, “T1” under goal one is Target 1.1, which is the eradication of extreme poverty for all people everywhere, currently measured as people living on less than USD 1.25 a day, by 2030. Another example is “T6” under goal one as Target 1.a (after Target 1.5), which is to ensure the significant mobilization of resources from a variety of sources, including through enhanced development cooperation, to provide adequate and predictable means for developing countries, in particular least-developed countries, to implement programs and policies to end poverty in all its dimensions (refer to the global goals link in Appendix A for all goals and their corresponding targets).

2.1. Selection of Delphi Experts

The assembly of the expert panel for the Delphi study on the impact of AI on achieving the SDGs underwent a meticulously crafted process, adhering to a systematic sequential sampling approach designed to ensure a comprehensive and scientific exploration of the subject matter. The selection criteria, deeply anchored in the qualifications and experience of professionals within the sustainability and SDGs’ sphere, emphasized not only their expertise but also their insights into the future implications of AI. This process of selecting experts for the Delphi panel, characterized by its flexibility and adaptability to this study’s specific research objectives [42], diligently evaluated expertise based on years of engagement in sustainability practices and a self-assessed understanding of the SDGs. A fundamental prerequisite for selection included a foundational knowledge of AI and its applications, ensuring that participants were well equipped to contribute substantively to this study. The impartial selection of participants was paramount, given this study’s emphasis on exploring forthcoming global advancements in AI and sustainability [40,43].

To achieve a fair and scientific selection process of experts, it was essential to have a clear, well-defined set of criteria. This included assessing potential panel members’ academic qualifications in sustainability-related fields, professional experience, established knowledge of sustainability practices, and active engagement in relevant sectors or academia [44]. This rigorous selection process aimed at assembling a diverse and competent group of professionals from various regions—including the United States, the United Kingdom, Singapore, Ghana, and South Africa—and was pivotal in cultivating a panel capable of providing a rich, multidimensional perspective on AI’s role in advancing the SDGs. The recruitment process, involving the distribution of invitations to 32 potential experts, with 15 expressing interest and ultimately 12 participating fully in this study, was conducted with attention on ensuring a broad representation of views and expertise. The selection process, aligning with established practices in Delphi methodology [44,45], enabled the formation of a panel that not only possessed deep insights into sustainability and AI but also represented a wide array of industries, further enriching this study’s findings. The careful curation of the expert panel, through a transparent and objective selection process, underscores this study’s commitment to scientific validity and reliability. The detailed criteria and systematic approach to expert recruitment ensure a balanced exploration of AI’s impact on the SDGs. Combined with the diversity and depth of expertise among the panelists, this approach contributes to an insightful analysis.

2.2. Delphi Cycles

The Delphi methodology, employed in this study to explore artificial intelligence’s (AI) role in achieving SDGs, is characterized by a structured, iterative process of panel consultations, designed to converge on expert consensus through multiple cycles. Historically, the effectiveness of a Delphi study hinges on the requisite number of cycles, with this investigation, reaching consensus after two planned cycles, each extending over a month [46]. This duration was deliberately chosen to provide panelists with sufficient time to offer in-depth contributions to the Delphi investigation. In the initial cycle, the research instrument was developed by integrating the relevant literature specific to AI’s contributions to the SDGs, ensuring a foundation grounded in current academic and practical understanding of the subject. The instrument for the subsequent cycle was then refined based on the insights and feedback obtained from experts during the first round, embodying the collective intelligence and foresight of the panel. This adaptive process allowed for the evolution of the questionnaire, making it a dynamic tool reflective of the panel’s growing insights.

Given the emergent and transformative nature of AI in the context of the SDGs, an open-ended question format was adopted, encouraging participants to articulate forward-thinking and innovative ideas freely. This approach facilitated the generation of a rich data set, which was analyzed using a qualitative thematic methodology, laying the groundwork for the iterative refinement of the research instrument in subsequent Delphi rounds. Throughout each cycle, expert perspectives were analyzed in detail, revealing a spectrum of consensus and divergence on various aspects of AI’s impact on the SDGs. The second cycle introduced a more structured questionnaire, enabling panelists to reflect on and evaluate the collective outcomes synthesized from the initial round’s discussions. The responses from this cycle were then subjected to a thorough analysis to confirm the emergence of consensus on the key topics under investigation. This iterative process, emblematic of the Delphi methodology, ensures that this study’s conclusions are robust, reflective of expert consensus, and grounded in a comprehensive understanding of AI’s potential and challenges in advancing the SDGs. The

Table 1. Background Information of Experts.

Demographic Designation	Number of Experts	Percentage
Academic qualifications
Bachelor’s degree	2	16.67%
Master’s degree	6	50.00%
Doctor of philosophy	4	33.33%
Total	12	100%
Area of specialization
Environmental sustainability and conservation, biodiversity, and climate action	3	25.00%
Urban planning and infrastructure	4	33.33%
Agriculture, health, and food security	2	16.67%
AI developers/scientists/enthusiasts	2	16.67%
AI data science and analytics	1	8.33%
Total	12	100%
Years of experience
1–5 years	2	16.67%
6–10 years	7	58.33%
11–15 years	2	16.67%
Over 15 years	1	8.33%
Total	12	100%
Employment agency
Academia	6	50.00%
Government agency	2	16.67%
NGO/international bodies/CSO	2	16.67%
Consultancy	1	8.33%
Private organization	1	8.33%
Total	12	100%

below shows the demographic of experts which participated in the study.

2.3. Achieving Consensus

Achieving consensus within a Delphi study, particularly when navigating a wide array of expert opinions on complex topics like the role of AI in advancing the SDGs, presents a challenge. The path to reaching consensus among panelists is not prescriptively outlined across Delphi studies, reflecting the dynamic and context-specific nature of this research approach. The concept of consensus, as highlighted by [47,48], is often viewed as a collective alignment or agreement that emerges from the panelists’ shared insights, encapsulating a mutual understanding or acknowledgment of commonalities within the diverse perspectives offered throughout the Delphi rounds. Historically, various statistical metrics have been employed to gauge consensus within the Delphi methodology. For instance, ref. [49] advocates for the use of mean item scores and standard deviation as indicators of consensus, positing that a narrowing of standard deviation across successive rounds typically signals a convergence towards collective agreement. Similarly, other scholars have incorporated the interquartile deviation (IQD) as a measure to assess consensus, adding another layer of analytical depth to the Delphi process [39,44].

In this study, a methodical application of IQD alongside mean item scores and standard deviation was utilized to determine consensus across Delphi rounds. The IQD metric calculates the spread between the 75th percentile (Q3) and the 25th percentile (Q1), indicating the central tendency dispersion among panelist responses. According to Aigbavboa [50], a lower IQD suggests a closer alignment of panelist perspectives, signifying a strong consensus. Conversely, a higher IQD reflects greater variability in opinions, indicating weaker consensus. The operationalization of these criteria for consensus assessment in the current study is detailed in

Table 2. Criteria for Assessing Consensus.

Status of Consensus	Interquartile Deviation (IQD)
Weak consensus	≥2.1 ≤3
Good consensus	≥1.1 ≤2
Strong consensus	≥0.0 ≤1

, delineating specific thresholds for weak, good, and strong consensus based on IQD values. This structured approach to evaluating consensus ensures a detailed and comprehensive analysis of panelist agreement levels. Such a methodology not only bolsters the validity and credibility of this study’s findings but also enriches the interpretive value of the insights gathered from the expert panel, providing a solid foundation for conclusions drawn regarding AI’s impact on the SDGs.

2.4. Determining Reliability and Validity of the Delphi Process

Ensuring the reliability and validity of the Delphi process is paramount for the integrity of its outcomes, particularly when exploring complex subjects like the intersections of AI with the SDGs. Reliability, as [51] delineates, refers to the consistency of a methodology in producing stable results under the same conditions. This notion becomes particularly evident within Delphi studies, categorized under qualitative research, where outcomes are heavily reliant on the subjective insights and perspectives of panelists [52]. The qualitative nature of Delphi studies, drawing upon the intuitive understanding and expert knowledge regarding AI’s contributions to the SDGs, introduces complexities in asserting reliability and validity. To address these challenges, a clear delineation of the Delphi process and its objectives was provided to the panelists, underscoring the importance of transparency in research communication. This study ensured that the objectives were comprehensively communicated to the participating AI and SDG experts.

This clarity was facilitated through detailed presentations of the targets and goals across all 17 SDGs, accompanied by potential applications and uses of AI within these contexts. Furthermore, experts were provided with a thorough guide on navigating the Delphi questionnaire, enhancing their understanding of this study’s scope and methodology. The selection of experts is critical to the validity of the Delphi process. According to [44], the expert selection phase functions as a crucial measure of validity within a Delphi study. Therefore, care was taken in the selection phase to ensure that participants were not only knowledgeable about AI and the SDGs but also represented a wide range of perspectives relevant to this study’s aims. This approach to expert selection aimed to encompass a diverse array of insights, thereby enriching this study’s data set and contributing to its overall validity. These measures collectively aimed to ensure that the insights generated from the Delphi study are both dependable and reflective of a broad, expert consensus on the role of AI in advancing the SDGs, providing a solid foundation for this study’s conclusions.

3. Results

3.1. Delphi Round One Result

The initial round of the Delphi study gathered responses from the expert panel on the impact of AI on achieving the SDGs. There was an opportunity for the inclusion of measures not initially captured in the survey by the experts. This study utilized a ten-point Likert scale, spanning from highly negative impact to highly positive impact. A summary of the results (showing the goals and the first three targets) of the first round of the Delphi study is presented in

Table 3. Results of Round One of the Delphi Study.

Goal/Targets (T)	Median	Mean		SD		IQD
Goal 1	8.00	8.00		0.70		1.000
T1	8.04	8.00		0.65		1.000
T2	6.67	6.67		0.49		0.500
T3	5.42	5.42		0.48		0.750
Goal 2	7.17	7.00		0.85		1.200
T1	8.00	8.21		0.63		1.000
T2	7.00	6.67		0.49		0.375
T3	5.50	5.42		0.48		0.375
Goal 3	8.00	8.00		0.70		1.000
T1	8.50	8.46		0.50		0.750
T2	7.00	6.71		0.47		1.000
T3	5.00	5.04		0.52		0.500
Goal 4	6.67	7.00		0.57		0.750
T1	8.50	8.46		0.50		0.375
T2	7.00	6.71		0.47		0.625
T3	5.00	5.04		0.52		0.750
Goal 5	7.08	7.00		0.64		1.000
T1	8.00	7.92		0.35		0.375
T2	5.75	5.75		0.42		0.375
T3	4.50	4.42		0.48		0.500
Goal 6	6.00	6.00		0.75		1.000
T1	8.00	7.92		0.35		0.500
T2	5.75	5.75		0.42		0.500
T3	4.50	4.42		0.48		0.500
Goal 7	7.83	8.00		0.78		1.250
T1	8.00	8.00		0.35		1.000
T2	6.50	6.42		0.42		1.000
T3	5.50	5.42		0.42		0.500
Goal 8	7.42	7.00		0.57		1.000
T1	8.00	8.00		0.35		1.000
T2	6.50	6.42		0.42		1.000
T3	5.50	5.42		0.42		0.500
Goal 9	8.17	8.00		0.79		1.000
T1	8.00	8.00		0.35		1.000
T2	6.50	6.42		0.42		0.500
T3	5.50	5.42		0.42		0.500
Goal 10	6.75	7.00		0.58		0.750
T1	8.00	8.00		0.35		1.000
T2	6.50	6.42		0.42		1.000
T3	5.50	5.42		0.42		1.000
Goal 11	7.42	7.50		0.63		1.000
T1	8.00	8.00		0.35		1.000
T2	6.50	6.42		0.42		1.000
T3	5.50	5.42		0.42		1.000
Goal 12	7.42	7.50		0.64		1.000
T1	8.00	8.00		0.35		1.000
T2	6.50	6.42		0.42		1.000
T3	5.50	5.42		0.42		1.000
Goal 13	6.42	6.50		0.50		0.750
T1	7.00	7.00		0.35		1.000
T2	7.50	7.67		0.42		0.500
T3	8.50	8.67		0.42		1.000
Goal 14	6.75	6.50		0.56		0.750
T1	7.00	7.00		0.35		1.000
T2	7.50	7.67		0.42		0.500
T3	8.50	8.67		0.42		1.000
Goal 15	7.83	8.00		0.62		1.000
T1	7.00	7.00		0.35		0.500
T2	7.50	7.67		0.42		0.500
T3	8.50	8.67		0.42		1.000
Goal 16	8.00	8.00		0.58		1.000
T1	7.00	7.00		0.35		0.500
T2	7.50	7.67		0.42		0.500
T3	8.50	8.67		0.42		0.000
Goal 17	7.75	8.00		0.50		1.000
T1	8.00	8.00		0.35		1.000
T2	6.50	6.42		0.42		0.500
T3	5.50	5.42		0.42		1.000
Cronbach alpha:	Goal: 0.89			Target: 0.96
Total goal contribution	Mean		Median		SD
All goal contributions	7.28		7.25		0.50
Cronbach alpha:	0.898

(See Appendix B.1 for the full table).

(see Appendix B.1 for the full table). This table showcases the findings of the mean, median, standard deviation, and IQD. The validity and reliability of the research instrument were ascertained using the Cronbach alpha test. This yielded a value of 0.96 for the targets, 0.89 for the goals, and 0.898 for the total contribution of AI, which suggests commendable validity and reliability of the research instrument for the Delphi study, as the value tends towards 1.00. The mean and median scores ranged between 8 and 10 for highly positive impact, 6 and 7.99 for positive impact, 5 and 5.99 for neutral impact, 3 and 4.99 for negative impact, and 1 and 2.99 for highly negative impact.

The mean scores and median scores assist in discovering the central tendency of the respondents. Hence, for the goal and target scores, which contained outliers from the views of the respondents, both the mean and median were used to determine the complete outlook of the perspectives of the respondents. In the context of AI’s impact on achieving the SDGs, the experts identified four goals with a highly positive impact of AI (very high score of the means: 8.00–10.00) goals. These include Goals 1, 3, 9, and 16. The lowest scores were within the neutral range, which included Goals 6, 10, 13, and 14. In the first round of the Delphi study, there was a strong consensus reached among the experts. The IQD results from the findings indicate that they were within the range of 0.3 and 1.5. However, due to some varying views and feedback, a second discussion was required to ensure a higher consensus. This study underscored the importance of understanding the influence of AI on each of the targets of the 17 goals and emphasized the need for continued dialogue and research in this domain.

3.2. Delphi Round Two Result

Table 4. Results of Round Two of the Delphi Study.

Goal/Targets (T)	Median	Mean		SD		IQD
Goal 1	7.05	7.00		0.32		0.600
T1	6.98	6.90		0.29		0.550
T2	7.02	7.02		0.28		0.530
T3	6.99	6.95		0.31		0.570
Goal 2	7.05	7.00		0.20		0.400
T1	7.05	7.04		0.22		0.410
T2	7.05	7.02		0.20		0.380
T3	7.05	7.02		0.20		0.375
Goal 3	7.04	7.00		0.21		0.350
T1	6.96	6.95		0.28		0.520
T2	6.97	6.95		0.26		0.500
T3	6.98	6.90		0.28		0.530
Goal 4	7.05	7.00		0.20		0.400
T1	7.01	7.00		0.29		0.530
T2	7.02	7.00		0.28		0.520
T3	7.00	6.99		0.29		0.530
Goal 5	7.00	6.98		0.15		0.150
T1	7.01	7.00		0.19		0.350
T2	7.00	7.00		0.14		0.305
T3	7.00	7.00		0.12		0.290
Goal 6	8.65	8.65		0.15		0.200
T1	8.61	8.60		0.16		0.250
T2	8.70	8.69		0.12		0.150
T3	8.63	8.60		0.13		0.200
Goal 7	8.92	8.90		0.16		0.200
T1	8.86	8.80		0.20		0.200
T2	8.90	8.89		0.15		0.200
T3	8.90	8.90		0.14		0.175
Goal 8	8.93	8.90		0.15		0.200
T1	8.75	8.75		0.12		0.150
T2	8.75	8.75		0.11		0.150
T3	8.77	8.75		0.10		0.150
Goal 9	8.95	8.93		0.18		0.225
T1	8.80	8.78		0.20		0.250
T2	8.80	8.77		0.18		0.200
T3	8.80	8.79		0.17		0.200
Goal 10	7.06	7.05		0.10		0.150
T1	7.00	6.98		0.21		0.275
T2	6.95	6.95		0.9		0.225
T3	6.95	6.95		0.17		0.225
Goal 11	8.97	8.95		0.15		0.225
T1	8.60	8.60		0.23		0.300
T2	8.60	8.60		0.22		0.250
T3	8.70	8.70		0.18		0.200
Goal 12	7.10	7.07		0.09		0.150
T1	6.96	6.95		0.14		0.200
T2	6.92	6.90		0.13		0.225
T3	6.93	6.90		0.13		0.175
Goal 13	8.82	8.82		0.01		0.010
T1	8.83	8.80		0.18		0.225
T2	8.82	8.80		0.15		0.200
T3	8.82	8.80		0.16		0.200
Goal 14	8.88	8.87		0.10		0.200
T1	8.73	8.70		0.20		0.250
T2	8.69	8.60		0.20		0.225
T3	8.72	8.70		0.17		0.200
Goal 15	8.93	8.90		0.16		0.200
T1	8.65	8.65		0.15		0.200
T2	8.60	8.58		0.12		0.175
T3	8.61	8.60		0.13		0.200
Goal 16	5.45	5.45		0.10		0.150
T1	5.42	5.40		0.09		0.150
T2	5.42	5.40		0.09		0.150
T3	5.41	5.40		0.08		0.150
Goal 17	5.53	5.50		0.07		0.200
T1	5.50	5.50		0.08		0.200
T2	5.40	5.40		0.10		0.200
T3	5.55	5.53		0.09		0.200
Cronbach alpha:	Goal: 0.91			Target: 0.97
Total goal contribution	Mean		Median		SD
All goal contributions	7.88		7.95		0.22
Cronbach alpha:	0.94

(See Appendix B.2 for the full table).

presents the summary findings from the second round of the Delphi study involving the experts who participated in the initial round (Table 4 contains the goals and the first three targets; see Appendix B.2 for the full table). After a careful deliberation of the responses to the first results, a second round of this study was conducted. This ensured a higher consensus of the expert views and perspectives.

The findings revealed that AI has a positive impact on the achievement of all 17 SDGs. Given the scale provided in Table 2, all 17 goals achieved good consensus (IQD values ranging from 0.01 to 1.0, based on the expert opinions in the Delphi study). Regarding AI’s impact on achieving the SDGs, the experts identified eight goals with a highly positive impact of AI (very high score of the means: 8.00–10.00) goals. These include Goals 6, 7, 8, 9, 11, 13, 14, and 15. The lowest scores were within the neutral range, which included Goals 16 and 17, respectively. The validity and reliability test for the research instrument further confirmed the robustness of the findings with an alpha value of 0.97 for the targets, 0.91 for the goals, and 0.94 indicating good reliability. The results according to Table 4 above underscores the positive impact of AI in achieving the SDGs at the goal and target level, emphasizing the need for the development of a comprehensive approach when integrating AI strategies with Sustainable Development Goals.

4. Discussion

AI is recognized for its transformative potential across a myriad of domains. Its influence on the United Nations’ Sustainable Development Goals is essential for both maximizing its benefits and addressing inherent challenges. Grounded in the Delphi approach, this study captures the expert consensus on the implications of AI for the SDGs. Through a Likert scale, this study delineates impacts, ranging from “highly positive impact” (8–10) to “highly negative impact” (1–2.99). The subsequent analysis thoroughly presents the median, mean, standard deviation (SD), and interquartile deviation (IQD) for each goal. This starts with Goal 1, which focuses on eradicating poverty globally. The sentiment towards this goal has a positive impact, with a median score of 7.05 and a mean of 7. This aligns with the belief that AI can transform our approach to combating poverty, from using AI in precision agriculture to improve food security to leveraging predictive analytics to identify regions vulnerable to economic challenges. The standard deviation of 0.32 and interquartile deviation of 0.6 indicate that while there is a consensus for a positive impact of AI’s role, experts are aware of its limitations and potential pitfalls.

For example, AI can optimize resource allocation, but relying too heavily on AI without human discretion may lead to unforeseen challenges [53]. Looking at specific targets within Goal 1, Target 1.1 (T1) shows that AI can play a significant role in ensuring equitable access to economic resources, with median and mean scores of 6.98 and 6.9, respectively [54]. This suggests that AI can greatly improve resource distribution, making essentials accessible to marginalized sectors. Similarly, Target 1.2 (T2), with scores of 7.02 for both the median and mean, indicates AI’s potential to strengthen the resilience of economically vulnerable populations, particularly in anticipating natural or economic adversities. Target 1.3 (T3) scores of 6.99 and 6.95 for the median and mean, respectively, highlight AI’s potential to enhance resource mobilization for anti-poverty strategies [7,53,54]. Overall, the data for Goal 1 demonstrate that AI can have a significant positive impact on poverty eradication, but there must be equity and fairness in its deployment. The experts’ insights suggest that AI’s applications, from precision agriculture to predictive analytics, have the potential to revolutionize how resources are distributed and how vulnerable populations are supported. For instance, the use of AI to enhance food security and to identify regions at risk of economic hardship was seen as a promising approach to tackling poverty at its roots. However, alongside these optimistic views, respondents also expressed concerns about the possible downsides of relying too heavily on AI solutions. They pointed out that without careful oversight, the deployment of AI might inadvertently widen existing social and economic gaps or lead to resource misallocation. Particularly in the context of Goal 1, there was a thorough discussion around ensuring that AI aids in making essential resources accessible to those in the margins, without sidelining human judgment and local knowledge.

Moving on to Goal 2, “No Hunger” and promoting sustainable agriculture, the data show a positive impact and emphasize the transformative potential of AI in sustainable development. This goal aims not only to feed the global population but to do so sustainably and efficiently. Examining the specific targets, Target 2.1 (T1) suggests that AI can play an instrumental role in guaranteeing access to nutritious food. Experts highlighted AI’s capacity to streamline food supply chains and reduce waste, thereby making nutritious food more accessible. Yet, they also underscored the importance of not undermining traditional agricultural practices or increasing smallholder farmers’ dependency on technology beyond their control or understanding. The expert panel emphasized the transformative role of AI in sustainable agriculture, particularly within the contexts of Target 2.2 (T2) and Target 2.3 (T3) of SDG 2. They agreed that AI-enhanced precision agriculture is a key advancement for optimizing resource use, achieving high yields with minimal environmental impact, and increasing the efficiency and sustainability of agricultural practices [55]. Moreover, the discussion on Target 2.3 revealed AI’s potential to simplify local market dynamics, providing producers with valuable insights into demand and helping align production more closely with consumption [16,55]. This application of AI is viewed as crucial for developing a more efficient, responsive agricultural sector informed by real-time data.

The panel’s insights further reveal artificial intelligence’s (AI) positive influence on achieving Goal 3, aimed at ensuring health and well-being for all, with AI’s application in diagnostics and predictive algorithms marking significant strides in global health. These technologies enable early intervention by healthcare professionals, notably improving maternal health outcomes and enhancing the precision in detecting and treating substance misuse, thereby offering more personalized and effective treatments [56]. In addressing Goal 3’s Target 3.3, experts discussed AI’s role in reducing traffic-related fatalities through its integration into urban planning and traffic management systems. AI’s capability to optimize mobility and decrease accidents underscores its potential to save lives through smarter city planning. The dialogue then transitioned to Goal 4, which focuses on inclusive and equitable quality education. The panel agreed on AI’s power to personalize learning experiences, ensuring that education adapts to the unique needs of each learner and aligns with the evolving demands of the job market, thus equipping students with relevant skills [57,58,59]. For Goal 5, centered on gender equality and the empowerment of women and girls, experts highlighted AI’s utility in identifying and mitigating discrimination, promoting equality through data analyses, and aiding in the proactive detection of harmful practices. This is seen as pivotal in fostering equitable representation in leadership and rectifying gender biases within organizations [60]. Finally, the discussion around Goal 6, dedicated to water and sanitation, emphasized AI’s critical role in enhancing the sector’s sustainability. The consensus on AI’s impact, reflected by a mean score of 8.65, illustrates its effectiveness in wastewater treatment and in preserving water ecosystems, with AI technologies ensuring the cleanliness of water sources and the efficient repurposing of waste [61,62]. This broad spectrum of AI applications across various SDGs showcases its transformative potential, urging a balanced and ethical approach to harness AI’s benefits while mitigating its risks. Furthermore, AI’s role in the preservation and restoration of water-related ecosystems underscores its comprehensive capabilities in contributing to sustainable water and sanitation management [61,62].

Within the context of Goal 6 on water and sanitation, experts underscored artificial intelligence’s (AI) pivotal role in both optimizing wastewater management and preserving water-related ecosystems. Their opinions reflect a consensus on AI’s comprehensive capabilities in enhancing sustainable water management, aligning with the observed scores and underscoring the potential of AI to contribute significantly to this area [61,62]. Regarding Goal 7, focused on ensuring access to affordable and sustainable energy, the panel’s insights and the high median and mean scores reveal a strong belief in AI’s transformative impact in the energy sector. Experts highlighted AI’s efficacy in augmenting renewable energy sources, optimizing energy usage through smart grids, and providing predictive maintenance and energy storage solutions, thereby affirming AI’s critical role in achieving sustainable energy access [63,64]. For Goal 8, which aims at sustainable economic growth and employment, the alignment between the experts’ views and the respective scores (mean and median scores of 8.9 and 8.93) emphasizes AI’s capabilities in forecasting job market trends, detecting forced labor, and ensuring workplace safety. The panel posited that AI is instrumental in propelling inclusive and sustainable economic advancement, echoing the quantitative findings [65].

In discussions on Goal 9, regarding resilient infrastructure, sustainable industrialization, and innovation, experts shared that AI’s transformative capacity is key to revolutionizing construction processes and streamlining industrial operations. The mean score of 8.93 illustrates the panel’s agreement on AI as a fundamental tool for fostering a resilient, sustainable, and inclusive future through innovation [7,66]. In addressing the reduction in inequalities (Goal 10), expert opinions resonated with the scores, particularly highlighting AI-driven financial tools (Target 10.1) for their potential in economic interventions. The experts concurred that AI has a significant role in analyzing financial trends and targeting interventions, thereby contributing to the reduction in economic disparities, and fostering inclusivity, thus substantiating the quantitative data presented [67]. For Targets 10.2 (T2) and 10.3 (T3), experts elaborated on AI’s vital monitoring capabilities that ensure policies are equitable and unbiased, and its significant role in managing global migration with policies that are both responsive and humane, illustrating AI’s potential to make policy formulation more inclusive and equitable [68]. This reflects the understanding surrounding AI’s capacity to enhance policy effectiveness and fairness in critical areas of social governance.

In the dialogue around Goal 11, the consensus on AI’s role in urban development was palpable. Experts pointed out AI’s impact on urban planning (Target 11.1), where AI technologies optimize space for equitable housing solutions, demonstrating AI’s ability to contribute to more inclusive and sustainable cities [69]. Similarly, AI’s application in traffic management (Target 11.2) and disaster management (Target 11.3) was highlighted for its potential to improve city living conditions by ensuring smoother traffic flow and enabling preemptive measures against disasters, thereby enhancing urban resilience [69]. Discussion on Goal 12 emphasized AI’s support for sustainable consumption and production patterns. AI’s optimization of resource extraction and utilization (Target 12.1), waste pattern prediction, and recycling process refinement (Target 12.2) were identified as key areas where AI contributes to more sustainable management practices. Additionally, AI-driven insights assist businesses in adopting sustainable strategies (Target 12.3), promoting a culture of responsible consumption [6,70]. In addressing Goal 13 on climate change, the consensus among experts highlights AI’s transformative role, acknowledging its predictive capabilities and real-time monitoring as vital for developing adaptive strategies and orchestrating effective global responses, thus enhancing climate resilience [71,72]. However, alongside the acknowledged mean score of 8.87, they also expressed concerns about the risks of overreliance on AI, such as data privacy breaches and potential misinterpretations of climatic data.

For Goal 14 on marine conservation, discussions reflected strong confidence in AI’s potential to maintain marine health and sustainability, supported by a mean score of 8.87. Experts praised AI for its efficiency in tracking pollution and facilitating sustainable fishing practices, crucial for preserving marine biodiversity and promoting the sustainable use of marine resources [32,73]. Yet, they also cautioned against unintended ecological impacts and the overlooking of community insights due to an overdependence on AI solutions in marine conservation efforts. For Goal 15, addressing terrestrial conservation, the consensus among experts, reflected in a mean score of 8.9 and a median of 8.93, showcased a highly favorable perception of AI’s role in this critical area. They pinpointed AI’s prowess in utilizing satellite imagery for the surveillance of unlawful activities such as logging, alongside its capacity to bolster reforestation projects and combat desertification. Such applications of AI are deemed pivotal in promoting the health of terrestrial ecosystems [32,74]. Turning to Goal 16, centered on peace, justice, and strong institutions, the response was more reserved, as evidenced by a mean score of 5.45. This score embodies a balanced assessment of AI’s capabilities, recognizing its utility in crime detection and prevention. Nonetheless, experts also raised concerns regarding the potential drawbacks of undue dependence on AI and the risks posed by algorithmic biases, which could potentially impair the principles of justice and equity [75,76]. The need for transparency in AI applications and the importance of utilizing impartial AI technologies in governance structures were underscored, highlighting the complex implications of AI’s integration into this sphere.

Regarding Goal 17, dedicated to bolstering global partnerships for sustainable development, a neutral mean score of 5.5 was observed, reflecting a measured appraisal of AI’s contributions. The expert panel acknowledged AI’s capacity to enhance international cooperation and streamline resource management and commercial transactions. However, they advocated for a strategic approach to AI’s deployment, emphasizing that it should augment, not replace, human initiatives [36,77]. This viewpoint articulates a cautiously positive perspective on employing AI to strengthen global partnerships, stressing the significance of achieving a symbiotic relationship between technological advancements and human endeavors. The SDGs encapsulate a global ambition to address pervasive challenges, with AI emerging as a crucial lever in this quest. Expert consensus recognizes AI’s capability to drive substantial progress in health (SDG3), education (SDG4), and the creation of resilient infrastructures (SDG9), highlighting its transformative potential [7,37]. In exploring the role of AI in achieving the SDGs, this study acknowledges the transformative potential of AI across various domains while also considering the complexities associated with its integration. Risks such as automation potentially exacerbating unemployment and deepening poverty in the technology and automotive sectors have been identified [16,78]. Concurrently, insights from the World Economic Forum and PwC underline AI’s crucial contributions to advancing Industry 4.0, particularly in industrial research, development innovation, sustainable infrastructure development, and the enhancement in IoT-enabled infrastructure efficiency [79,80]. This alignment with Goal 11 highlights AI’s significant utility in fostering sustainable urban and community development, presenting a landscape of AI’s potential to drive progress across the SDGs. To enrich the discussion further, experts explored the synergies and potential trade-offs between AI impacts across various SDGs. For example, AI’s role in enhancing precision agriculture (Goal 2: Zero Hunger) directly contributes to food security while also supporting Goal 13: Climate Action by optimizing resource use and reducing waste.

Similarly, AI’s impact on improving access to quality education (Goal 4) is intricately linked to fostering innovation and economic growth (Goal 8), preparing individuals with the skills required for the future job market, including those in the AI and sustainability sectors. However, alongside these positive contributions, the revised discussion also critically examines the challenges highlighted by experts. Ethical considerations in AI deployment and the imperative to ensure AI technologies are inclusive and do not exacerbate existing inequalities (Goals 5 and 10) are emphasized. This holistic perspective underscores the importance of a balanced and ethical approach to AI integration across all SDGs, acknowledging that advancements in one area may influence outcomes in others. The dialogue among experts culminates in a call for a delicate approach to AI’s integration within the SDGs’ framework, advocating for strategies that are sustainable, inclusive, and equitable. Concerns regarding the potential for an unregulated deployment of AI to widen the digital divide underscore the need for ethical deployment practices. In their conclusive assessment, respondents reached a consensus indicating a predominantly positive impact of AI on achieving the SDGs, with an average impact score of 7.88 out of 10. Such consensus aligns with analyses positing AI as an enabler of most SDGs, facilitating advancements in approximately 79% of them [7]. This collective evaluation by experts presents a balanced examination of AI’s capabilities and limitations, urging a careful application of AI technologies. These insights underscore the necessity of responsibly exploiting AI’s potential to ensure that its integration into efforts to achieve the SDGs contributes to a sustainable and inclusive future for all.

5. Conclusions

This research aimed to elucidate the complex role of AI in supporting the SDGs. By employing a Delphi methodology, this study engaged a panel of distinguished experts to assess AI’s diverse impacts across various sectors, juxtaposing its extensive benefits against the inherent challenges within the context of sustainable development. The meticulous evaluation utilized a Likert scale to quantify impacts from “highly positive impact” to “highly negative impact,” yielding a detailed understanding of AI’s contributions and implications for the SDGs. The analysis revealed AI’s significant transformative potential, particularly evident in efforts to eradicate global poverty (Goal 1) and enhance marine conservation (Goal 14). Conversely, a more nuanced view emerged on its role in promoting peaceful and inclusive societies (Goal 16), underscoring the necessity for a strategic, balanced, and ethical approach to integrating AI with sustainability efforts. Derived metrics, including the median, mean, standard deviation (SD), and interquartile deviation (IQD), highlighted a consensus among experts on AI’s critical role in advancing toward the SDGs, especially noted in urban design, traffic optimization, and disaster response. These areas align with scholarly contributions that emphasize technology’s pivotal role in sustainable urban development [69].

Consequently, this investigation underscores the dual nature of AI’s impact on sustainable development, advocating for the formulation of policies, strategies, and frameworks that harness AI’s capabilities responsibly. The imperative for ethical considerations and a commitment to the well-being of all stakeholders is paramount in leveraging AI to address the challenges posed by the SDGs. In understanding these insights, the conclusion of this study presents a comprehensive summary of AI’s instrumental role in achieving the SDGs. It becomes evident that AI, when employed strategically and ethically, emerges as a formidable ally in the quest for sustainability. However, the integration of AI in efforts to achieve a sustainable and inclusive future necessitates a commitment to strategic foresight and ethical principles, ensuring that technological advancements serve as catalysts for positive change across all SDG domains. Therefore, this study advocates for an integrated approach to AI implementation, one that harmonizes technological advancements with the foundational values and objectives of the SDGs. This research calls for collaboration among policymakers, practitioners, and technologists to develop a framework that prioritizes ethical deployment and maximizes AI’s potential for sustainable development, thereby reinforcing AI’s role as a pivotal tool in the global pursuit of sustainability.

Limitations and Implications for Future Research

In the spirit of scholarly precision, it is crucial to recognize the limitations of this study, which subsequently highlights avenues for further research. This study’s exploration into the impacts of AI on the SDGs could be enhanced through a more exhaustive and systematic assessment of the relevant literature. Consequently, future investigations should embark on a deeper exploration of the determinants through a comprehensive literature review. The organization of determinants could be optimized by categorizing them into three primary AI dimensions, technological, ethical, and societal, thereby more accurately reflecting the dynamic landscape of AI research. Future studies could benefit from the application of a principal component analysis, based on these categories, to offer a more quantitative viewpoint on AI’s contributions to the SDGs. In this study, a diverse group of experts was assembled, effectively limiting biases and gathering varied insights on the impact of artificial intelligence on sustainable development. Despite these strengths, the range of perspectives remained somewhat restricted by the participant count. It is recognized that an expansion in the array of viewpoints could enhance the depth of future research. Thus, it is recommended for subsequent studies to not only preserve the diversity of expertise but also to augment the number of experts involved. An increase in participant numbers is anticipated to enrich the consensus process, further solidify the study’s validity, and ensure a comprehensive exploration of artificial intelligence’s role in achieving the SDGs. Such an approach is poised to elevate the empirical rigor and make a significant contribution to the field. While the present study emphasized consensus through the interquartile deviation (IQD) method, the incorporation of other statistical techniques, such as the coefficient of variation (CV), could provide a more definitive consensus. A comparative study of the efficacy of both CV and IQD methods within a Delphi framework, specifically focusing on AI and the SDGs, would be invaluable in enhancing the understanding of Delphi methodologies. Furthermore, future research should consider the conversion of the identified determinants into indicators, thereby facilitating the development of models that predict the impact of AI on SDG outcomes. These determinants could also serve as the foundation for a policy framework dedicated to AI’s role in achieving the SDGs. In practical terms, such a framework would be crucial for nations aiming to align their AI strategies with sustainable development objectives, a task of utmost importance in navigating the complex challenges of the contemporary era, with AI positioned as a key catalyst for change and progress.