Artificial Intelligence in Project Success: A Systematic Literature Review

Abstract

Projects play a vital role in achieving organizational success, where artificial intelligence (AI) has a transforming impact in project management (PM). The integration of AI techniques into PM practices has the potential to significantly improve project success rates and enable more effective project management. This article adopted a systematic literature review (SLR) methodology, following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and employing a content analysis strategy to review 61 peer-reviewed academic journal articles published between 2015 and 2025 in the Web of Science and Scopus. This study investigates the key project success dimensions influenced by AI throughout the project lifecycle, and identifies the AI sub-fields and algorithms employed in relation to project success, where time and cost are found to be the most significantly affected factors in project success. Machine learning (ML), along with its corresponding algorithms, emerged as the most frequently applied AI subfield. This study overviews key AI-influenced project success factors and the main AI subfields and algorithms in recent literature, providing actionable insights for diverse project stakeholders aiming to enhance outcomes through AI. Limitations, including the lack of industry or regional focus, exclusion of project management process groups, and omission of gray literature, were also acknowledged, which suggest valuable directions for future research.

1. Introduction

The generation of artificial intelligence (AI) can be traced back to the groundbreaking ideas of Vannevar Bush called memex, which is a machine envisioned as an expansive, intimate extension of human memory [1]. The terminology “artificial intelligence” was first introduced by John McCarthy in his proposal for the Dartmouth Summer Research Project in 1955 [2,3]. Subsequently, in the 1960s, early programs such as ELIZA and the General Problem Solver were developed based on the belief that human intelligence could be formally represented [4].

The rise of AI is revolutionizing traditional business models, challenging conventional methods, and ushering in a new era of digital transformation [5,6]. These advanced AI systems, capable of learning, adapting, and generating content, redefine innovation by allowing companies to automate complex creative processes that once required significant human intelligence and effort [7]. In this dynamic landscape, AI technology transcends the boundaries of routine operational tasks [8], making its mark on the intellectual domain as well. The primary goal of AI is to develop technologies that enhance human intelligence or assume tasks that are tedious, repetitive, hazardous, demeaning, or dehumanizing [9]. Through generative AI tools like ChatGPT, employees can now access a unique source of creative potential [7], enabling them to perform tasks traditionally managed by humans [10]. AI can also address the limitations of human information processing, offering advanced capabilities to identify issues, opportunities, and risks that go beyond conventional search methods and knowledge boundaries [11]. These advances improve project-related activities such as planning, risk assessment, and decision making. Common and rapidly advancing subfields of AI research include machine learning, natural language processing, expert systems, among others [12].

Given AI’s growing role, it is important to clarify what defines a project and how its success is measured. Kerzner [13] describes a project as a singular and time-limited initiative aimed at producing a particular product, service, or outcome with defined goals, time-frames, and allocated resources. Every project is unique and comes with its own set of risks, which influence the success of the organization over time with changes in tasks and team members. Since the 1960s, researchers in project management (PM) have been exploring the factors that contribute to project success [14]. A project is considered successful when its objectives are achieved [15] and is traditionally evaluated based on time, cost, and quality [16], which are readily quantifiable. Baker et al. [17] introduced client satisfaction as an additional factor to the conventional project success criteria, including the Iron triangle of cost, time, and quality. Consequently, the criteria for project success evolved into a square that includes cost, time, quality, and client satisfaction [18]. In addition, Okudan et al. [19] highlighted risk management (RM) as a critical success factor for project success and company operation. Meanwhile, Zabin et al. [20] also underscored that informed decision-making would facilitate project outcomes. The aforementioned points represent critical success factors for project success.

AI is increasingly applied across domains, especially in the rapidly growing field of PM [21]. Niu et al. [22] emphasized the substantial potential of AI in PM, enabling project managers to improve the accuracy, precision, and speed of decision-making processes. Typical human errors that frequently lead to project failures can be mitigated through the integration of AI techniques across various phases of project execution. By enhancing forecasting capabilities and addressing both administrative and technical inaccuracies, AI contributes to more robust and data-driven decision-making processes—an imperative development given that current decision-making frameworks in the construction industry remain predominantly human-centric. However, such human-centric models often fail to fully recognize the potential of AI to process information with greater accuracy, precision, and speed, thereby enabling superior decision-making under complex or uncertain conditions.

Niu et al. [22] use smart construction objects as an example of AI-enabled construction resources. These include machinery, tools, or materials embedded with sensing, processing, and communication capabilities. Informed by ubiquitous computing theories, smart construction objects are equipped with autonomy and contextual awareness, allowing real-time environmental interaction and delivering adaptive, data-driven insights that enhance the speed and accuracy of decision-making, with or without human intervention. What distinguishes smart construction objects from conventional construction objects is their ability to communicate directly with each other. For instance, a smart tower crane can exchange information with smart construction materials to evaluate potential safety risks before lifting operations. This is particularly beneficial for managing uncertainty, improving efficiency, optimizing schedules, engaging stakeholders, and mitigating risks. Furthermore, as noted by Bushuyev et al. [23], the use of AI-based analytics enables project managers to more effectively address challenges such as risk mitigation, flexible resource allocation, and stakeholder engagement, thereby leading to more robust and adaptable project outcomes.

The integration of AI is particularly beneficial for projects characterized by high levels of uncertainty, limited resources, and complex risk factors. In addition, AI has the potential to reduce both the time and cost associated with physical prototyping, optimize resource management to ensure a more efficient utilization of labor and materials, and automate routine tasks, allowing human personnel to focus on creative and strategic activities [24]. Its predictive modeling capabilities enable the anticipation of potential problems and obstacles in the innovation process, thus accelerating the time to market [25]. Moreover, recognizing the importance of customer satisfaction, AI facilitates product customization to better align with individual customer needs [26].

While numerous studies have investigated the application of various AI algorithms to enhance certain aspects of project success, most existing research remains largely technical and experimental in nature [23,27,28]. There is a notable lack of comprehensive systematic literature reviews synthesizing these fragmented findings to provide a holistic understanding of AI’s impact on project success. This gap limits the ability of researchers and practitioners to discern overarching patterns, identify consistent benefits, and uncover areas requiring further exploration. Beyond the scarcity of systematic reviews, existing studies predominantly focus on specific AI algorithms applied to isolated aspects of PM, such as risk prediction [19] or decision-making support [29]. However, a comprehensive understanding of how AI influences multiple dimensions of project success—including time, cost, quality, stakeholder management, and decision making—is lacking. Furthermore, most existing studies tend to focus on the application of AI in PM within a single field, such as construction projects [30], software projects [31], or manufacturing projects [9]. However, cross-industry comparative or integrative analyses remain scarce, limiting the generalizability of findings and the understanding of how AI may differently impact project success across various sectors.

To further elaborate on these trends, several representative studies are presented below to illustrate how AI has been utilized in PM and what outcomes have been reported. Glahe et al. [32] developed a no-code web platform based on a machine learning algorithm, which enables us to improve the working efficiency, including time, cost, and quality aspects in an agriculture project. While Fernandes et al. [29] created a practical AI chatbot application underpinned with a natural language processing algorithm for construction project usage, which can be used in project time and decision-making improvement. In a construction project, Karki and Hadikusumo [33] proposed a project manager’s competency prediction model based on a machine learning approach, which is important for a project’s decision-making support. Borrero-Domínguez and Escobar-Rodríguez [34] also proposed a decision support system in a crowdfunding project based on a fuzzy cognitive maps approach, shown in stakeholder management improvement. A machine learning and natural language processing-based clinical system is created in [35], resulting in time and cost savings in clinical projects. Moreover, Mitrovic et al. [36] developed an artificial neural network, machine learning project–outcome prediction model for project decision-making in software projects. In addition, Bushuyev et al. [23] propose a conceptual framework for applying AI to sustainable development projects to support the entire project success.

As discussed above, existing studies are fragmented—typically addressing isolated project dimensions, applying single AI techniques, and limited to specific industry contexts. Despite growing interest in AI’s potential to enhance project success, the literature remains predominantly technical and context-bound, with a lack of comprehensive integrative reviews. To address this gap, this study conducts a systematic literature review (SLR) with the aim of consolidating current research and identifying two major questions in the application of AI to project success. Specifically, the review seeks to answer the following research questions:

Research Question 1: what are the key project success dimensions influenced by AI throughout the project lifecycle?

Research Question 2: what AI sub-fields and algorithms have been employed in relation to project success?

2. Methodology

A systematic review involves a comprehensive search of academic databases, journals, conference proceedings, and other sources to find relevant studies. For this review, Web of Science and Scopus, two of the largest scientific literature databases, were used as the main sources.

This review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement [37] (see Supplementary Materials). PRISMA improves systematic reviews by helping authors clearly report why, how, and what was found. This ensures reviews are transparent, complete, and accurate, supporting evidence-based decisions [37]. The overall research process consists of four key steps: identification, screening, eligibility, and inclusion for review.

This SLR ground research is restricted to information concerning the use of AI in project success and the available databases. The keywords for the search were derived from the research objective, which aims to investigate studies of how AI addresses various aspects of project success. Hence, the main keywords used for the search were “artificial intelligence and project success” (Topic) or “AI and project success” (Topic). The word (OR) is used as a conjunction, resulting in more focused and productive results. An initial search conducted in the Web of Science database identified 288 records. The inclusion criteria were restricted to peer-reviewed journal articles written in English and published within the last ten years (2015–2025), details in

Table 1. Inclusion and exclusion criteria.

Criterion	Inclusion	Exclusion
Literature Type	Peer-reviewed journal articles	Non-peer-reviewed journal articles
Language	English	Non-English
Timeline	2015–2025	Before 2015
Accessibility	Papers can be retrieved	Papers cannot be retrieved

below, as such articles reflect both academic rigor and the current state of the art in the field [38]. Additionally, a search using the same keywords in the Scopus database yielded 375 records, with the same restrictions applied to document type, language, and publication period (2015–2025). The exclusion process was conducted in accordance with the PRISMA 2020 guidelines. During the identification phase, 180 duplicate records were removed. Subsequently, 375 records were excluded based on title and abstract screening due to irrelevance to the research topic.

In the screening phase, 13 records were excluded because the full text could not be retrieved. While this may have had a minor impact on the completeness of the review, these records made up only a small portion of the initial dataset and were removed before the final selection of included studies. We believe their exclusion does not significantly affect the reliability of the findings or the key conclusions, as the included studies sufficiently cover the core aspects of the research questions.

At the eligibility stage, additional records were removed because they were either not empirical research or were not aligned with the focus of the study after full-text assessment. As a result, a total of 61 studies were included in the final systematic review. During the process, data was extracted by the author, and then the accuracy was checked by the coauthor. The detailed PRISMA process flow diagram is provided in Figure 1.

3. Results

In this section, the statistical results for both the involved project success factors and the AI sub-fields and algorithms are presented based on the PRISMA analysis.

3.1. Identification of Project Success Factors

The analysis of the selected studies indicates that AI technologies influence multiple project success factors throughout the project lifecycle. Specifically, the literature consistently identifies AI’s impact on key success factors such as time, cost, quality assurance, risk management, stakeholder management, decision-making support, and project success in general. Among the identified dimensions of AI impact on project success, time and cost emerge as the most frequently reported factors, appearing in 21 and 20 studies respectively, indicating a strong consensus within the literature regarding their critical importance. The decision-making support dimension, a multi-faceted construct comprising 13 sub-factors (detailed in

Table 2. Identified project success dimensions and sub-level factors.

Dimension	Frequency	Sub-Level Dimensions
Time	21	N/A
Cost	20	N/A
Quality	11	N/A
Risk management	6	N/A
Decision-making support	19	Critical success factors; technology selection; prediction models; performance prediction; priority setting; resource allocation; trade-off modeling; efficiency indicators; expert integration; project selection; HR assigning; incident prediction
Stakeholder management	4	N/A
Project success in general	13	New product development (NPD) process; Knowledge management; Innovation improvement; Firm and business performance; Process automation; Agility; Employee competency; Holistic views

), is referenced in 19 studies. Similarly, project success in general, also multi-dimensional (see Table 2), is reported in 13 studies. Other factors, including quality, risk management, and stakeholder management, appeared less frequently, cited in 11, 6, and 4 studies, respectively.

3.2. AI Sub-Fields and Algorithm Identification

A diverse range of AI sub-fields have been employed in the literature addressing project success, with applications observed across multiple industries, including but not limited to construction, information technology, agriculture, and engineering. In this reviewing process, a total of six types of AI sub-fields were identified across the 61 reviewed studies. The most prevalent category was machine learning (ML), utilized in over half of the reviewed papers (33 studies), where algorithms such as deep learning (DL), support vector machines (SVM), artificial neural networks (ANN), and other related algorithms (see

Table 3. AI sub-fields and associated algorithms in reviewed studies.

AI Sub-Field	Frequency	Algorithms (If Applicable)
Natural Language Processing (NLP)	4	Generative Pre-Trained Transformer
Machine Learning (ML)	33	Deep Learning; SVM; Random Forest; Genetic-SVM; Genetic-RF; ANN; LSTM; GRU; DNNs; Lasso; Ridge; CNNs; Lazy Learning; LLMs; Decision Tree; GLM
Knowledge Representation and Reasoning (KRR)	5	Case-Based Reasoning; Bayesian Fusion; Symbolic AI; AHP; TOPSIS
Computational Intelligence (CI)	4	Fuzzy Logic; Fuzzy Clustering; Flow Direction Algorithm; Max–Min Ant System
Hybrid Intelligent Systems (HIS)	6	LLNF; Fuzzy Expert System; Swarm Intelligence; SFNE; FBWM; FCM
Not specific (General AI)	17	N/A

) were mentioned in different studies, followed by 17 studies that discussed AI in general. Hybrid Intelligent Systems were applied in six studies, with five studies in knowledge representation and reasoning (KRR), while both natural language processing (NLP) and computational intelligence (CI) were the least frequently reported, each appearing in only four studies. Under each AI sub-field, there exists a set of specialized algorithms tailored to its unique objectives, which are also listed in Table 3. These results highlight the dominance of data-driven approaches such as ML, while AI in general takes a secondary position (See Figure 2), with the research result also indicating a smaller but diverse application of other AI methods within the context of project success.

4. Discussion

This section addresses each of the two research questions outlined in the introduction, providing insights into how the literature contributes to answering them. To ensure coherence, the discussion follows the sequence of the research questions as outlined in the introduction. The study has used typical cases from reviewed articles to yield more nuanced answers to the research questions. A comprehensive list of these typical cases from various project domains is provided in

Table 4. AI applications in different project contexts.

Reference	Year	Project Domain	AI Sub-Field	Project Success Dimension
[27]	2022	Construction project	Machine Learning	Decision-making support
[19]	2021	Construction project	Knowledge Representation and Reasoning, Machine Learning	Risk management
[30]	2021	Construction project	AI in general	Time, cost, quality
[28]	2025	Construction project	Machine Learning, Computational Intelligence	Quality and decision-making support
[40]	2024	Construction project	Knowledge Representation and Reasoning, Machine Learning	Cost
[29]	2024	Construction project	Natural Language Processing	Time and decision-making support
[41]	2022	Construction project	Machine Learning	Cost and decision-making support
[42]	2024	Construction project	Computational Intelligence	Decision-making support
[34]	2023	Crowdfunding project	Hybrid Intelligent System	Stakeholders management
[43]	2023	Energy project	Machine Learning	Time, cost and decision-making support
[44]	2021	Not specific	Natural Language Processing, Machine Learning	Time, Project success in general
[45]	2020	Not specific	Machine Learning	Decision-making supporting
[46]	2023	Healthcare project	Machine Learning	Stakeholders management
[35]	2022	Healthcare project	Natural Language Processing, Machine Learning	Time and cost
[47]	2021	High-tech project	AI in general	Project success in general
[9]	2022	Manufacturing project	Machine Learning	Quality
[48]	2020	Manufacturing project	Machine Learning	Time, cost, quality
[49]	2023	Organizational behavior and management project	Machine Learning	Project success in general
[50]	2015	Software project	Hybrid Intelligent System	Time and cost
[51]	2018	Software project	Knowledge Representation and Reasoning, Machine Learning	Decision-making supporting
[52]	2015	Software project	Computational Intelligence	Time and cost
[36]	2020	Software project	Machine Learning	Decision-making support
[31]	2025	Software project	Hybrid Intelligent System	Cost
[53]	2024	Space project	Machine Learning	Risk and cost

, which shows that construction projects predominate among the typical examples and across all reviewed studies, followed by software projects.

4.1. Research Question 1: What Are the Key Project Success Dimensions Influenced by AI Throughout the Project Lifecycle?

After a comprehensive review of the selected studies, seven key project success factors were found to be significantly influenced by the application of AI technologies. These factors include decision-making, project success, risk management, time, cost, quality, and stakeholder management.

4.1.1. “Iron Triangle” Time, Cost, and Quality

Time, cost, and quality are defined as the iron triangle in tradition project management [27]. The following studies fully or partially cover the project success dimension within the iron triangle. In the study by [30], the authors posit that AI has the potential to enhance project success in terms of cost, time, and quality, thereby increasing the likelihood of project success. However, the research suggests that the significance of AI as a crucial factor in the success of construction projects has not yet been thoroughly investigated.

At present, AI remains at a nascent stage in the context of construction project execution. Its application is largely confined to design computations and the support of project managers in performing routine, repetitive tasks. The primary limitations of AI, as identified by our interview respondents, include its lack of soft skills, the absence of human-like cognitive abilities necessary for interpreting situations from multiple perspectives, its inability to cultivate interpersonal relationships, and the way humans manage projects. Escobar et al. [9] centered their proposal on the project quality dimension, which motivated manufacturing and quality engineers to be able to drive AI-based continuous improvement and short-term innovation using the seven-step problem-solving strategy for Quality 4.0. Their study employed a real-world case practice, systematically applying each step of the seven-step problem-solving methodology to address a complex quality-related issue. Furthermore, Fan [28] stated that two soft computing techniques are integrated to uncover intricate yet meaningful relationships among various factors affecting construction, thereby facilitating predictive analysis without exclusive dependence on expert-based models. Initially, significant construction factors with higher weights are pinpointed through fuzzy logic assessment. Subsequently, an ANN based AI model is applied to model the relationships between these critical factors and construction quality. The outcomes of these predictions can guide construction management approaches, enhancing construction quality and overall project success. The developed prediction model was further evaluated to examine the relationship between the identified key factors and construction quality in this study, yielding a prediction accuracy of 96.08 percent.

Long and Anh [40] proposed AI methods that demonstrate superior accuracy compared to conventional techniques currently employed by construction companies in Vietnam. This case-based reasoning (CBR) AI model outperforms in terms of its description of how building costs are estimated, bringing benefits to building cost calculations. However, this method faces challenges when applied to complex projects, as transferring case-specific elements across different scenarios can be intricate and may limit its broader applicability.

4.1.2. Risk Management

Risk management is a series of proactive efforts aimed at increasing the probability and impact of positive risks while reducing those of negative risks. Effective risk management enhances project success by increasing the likelihood of positive risks and minimizing the impact of negative ones [19]. From the perspective of dynamic capability theory, internal dynamic IT capabilities such as AI can be used as an effective tool for sensing, which involves identifying and evaluating potential opportunities and risks [54].

Okudan et al. [19] highlighted that implementing continuous risk management (RM) throughout the life cycle of the project can help decision-makers to formulate proactive risk response strategies that emerge during a project. Given the fact that prevention is always better than a cure, proactive response strategies are certainly the key to achieving project objectives. To support this approach, they developed a knowledge-based RM tool using AI technology, which aims to improve the efficiency of RM in construction projects and can be adapted for use in other project-based sectors with minor adjustments. In the research conducted by [53], a methodology for the intelligent evaluation of effectiveness and risk assessment of investments in high-risk start-ups was introduced, particularly within the space sector. The proposed AI framework integrates multiple established AI-driven techniques in concert, enabling a robust and comprehensive evaluation of both investment effectiveness and associated risks within the context of the space industry.

4.1.3. Decision-Making Support

Decision-making is also closely linked to dynamic IT capabilities, as the effective use of AI enables enterprises to make informed decisions when seizing opportunities and to implement strategic transformations that support sustained competitive advantage [54]. Effective decision-making is critical to project success, and AI-based decision support systems provide insightful recommendations by leveraging data analytics [34].

Bang et al. [27] suggested that employing machine learning techniques to identify critical success factors is feasible, with the model demonstrating strong decision-making performance. Many of the key features identified relate to the project’s initial stages, including planning, analysis, and engineering. This finding implies that project success can potentially be predicted early by evaluating, reporting, and monitoring these features during the initial phases. The suggested AI model effectively uncovers significant factors contributing to project success, which can be leveraged in early project stages to forecast outcomes in later phases or for the project as a whole. Consequently, such evaluations may serve as valuable tools to enhance the likelihood of achieving greater project success. In [36], the authors also proposed that the utilization of critical success factors can facilitate the development of predictive models that deliver important insights to project managers and enhance software development projects and software project success. Unlike traditional parametric methods that rely on historical project data to formulate linear models through regression analysis, the proposed AI technique is not limited by data distribution assumptions and does not require a predefined model structure [55].

4.1.4. Stakeholder Management

From the stakeholder management perspective, the fundamental premise of stakeholder theory is that the involvement and active participation of stakeholders can significantly enhance organizational performance, and the satisfaction of customers constitutes a fundamental determinant of project success [56].

Zaidi et al. [46] developed an intelligent AI-driven approach leveraging machine learning algorithms to predict and detect cybersickness. This system enables better stakeholder management by anticipating different severity levels of CS in users of virtual reality technology. The most notable contribution of [34] lies in identifying guidelines for developing an innovation AI system based on a Fuzzy Cognitive Maps (FCMs) approach, aimed at predicting the success of crowdfunding projects based on their profiles. These findings are significant to key stakeholders like the promoter and funder. The promoters can utilize the elements discussed in the article to enhance the likelihood of project success. Regarding the funder, they can have the opportunity to invest in projects that are most suitable and have the highest potential for success.

4.1.5. Project Success in General

Last but not least, AI plays a role in facilitating project success in general. Zhang et al. [47] proposed that organizations with greater new product development (NPD) success show greater organizational commitment to AI and invest more in AI capability. Embracing, promoting, and investing in AI capability increases NPD success rate and reduces failure rate. Actively embracing, promoting, and investing in AI contributes to higher NPD success rates and a reduction in failure rates. Despite AI’s significant potential to enhance NPD outcomes, its adoption within the NPD process remains relatively limited, with only 49 percent of organizations reporting its use. The study claims that its proposed best practices could be used to effectively integrate AI technologies into NPD processes and enhance overall project success. While in the study conducted by [49], through the power of AI, the study proposed an AI model which can predict the organizational agility level of a company. This model integrated prevalent agile principles and practices, while also taking into account the organization’s size and type, as well as the supportive structures that are crucial for overall project success.

4.2. Summary of Findings for Q1

Throughout the reviewed literature, it has been demonstrated that AI technologies can significantly contribute to project success from multiple perspectives. For example, they can improve organizations’ decision-making processes, especially during the initial phases of projects [27,36]. Furthermore, the adoption of AI is strongly linked to enhanced project success indicators such as cost, time, and quality [30,34,46] studies that emphasized the positive impact of AI on stakeholder management as one of the project success factors. Nonetheless, the extent of AI integration differs among industries and project categories, with areas like new product development [47] and construction [19,30] revealing more emerging yet promising uses. In conclusion, AI technologies are extensively used to facilitate different project success dimensions. However, the maturity level in different industries varied.

4.3. Research Question 2: What AI Sub-Fields and Algorithms Have Been Employed in Relation to Project Success?

Through the selected literature, five AI sub-fields were identified, with multiple AI algorithms under each subfield. The AI sub-fields are natural language processing (NLP), machine learning (ML), knowledge representation and reasoning (KRR), computational intelligence (CI), and hybrid intelligent system (HIS).

4.3.1. Natural Language Processing (NLP)

Natural language processing (NLP), a branch of AI, leverages computational algorithms and statistical techniques to analyze unstructured text and extract linguistic patterns from human language [57]. Fernandes et al. [29] leveraged GPT-driven natural language processing to develop the Digital Assistant for Virtual Engineering (DAVE), which offers an intuitive and efficient way for users to interact with complex Building Information Modeling environments via both text and voice interfaces. DAVE exemplifies how advanced AI technologies can enhance operational efficiency and facilitate more informed decision-making in construction projects by making BIM systems more accessible to a broader user base. In the study by [35], clinical decision support tools were developed by integrating two AI technologies: machine learning and natural language processing. These tools were implemented in real time to provide support at natural points in the clinical workflow, minimizing disruptions and avoiding delays in patient care. However, to ensure continued effectiveness, the system requires alignment with industry standards, automated data access, seamless integration into clinical workflows, and ongoing user feedback.

4.3.2. Machine Learning (ML)

Another AI technique, machine learning (ML), is a subset of AI that is trained on data to make human-like decisions [44]. Machine learning enables computer systems to process and learn from real-world data, thereby improving their performance on specific tasks through experience [58]. While these systems imitate aspects of human learning, their learning process is driven by algorithmic pattern recognition rather than human reasoning. Sabahi and Parast [45] used multiple machine learning algorithms, including lasso regression, ridge regression, support vector machines, neural networks, and random forests to investigate the relationship between individuals’ project performance and measurements of entrepreneurial orientation and entrepreneurial attitude. The results showed that the best method for predicting project performance among the machine learning algorithms is lasso regression. Roslon [41] also proposed that artificial neural networks, an algorithm of machine learning, can support the selection of optimal material or technological solutions. These tools assist decision-makers and ultimately contribute to improved project outcomes. Silveira et al. [48] presented a flexible pilot project methodology by leveraging machine learning techniques, along with a deployment example in the context of a semiconductor company. The authors also emphasized the importance of data for machine learning algorithms, which must be provided with sufficient and reliable input in order to effectively control the monitored variables. These variables include those used in prediction algorithms, statistical process control, data compression, and software sensor applications.

4.3.3. Knowledge Representation and Reasoning (KRR)

Knowledge representation and reasoning (KRR) is another sub-field of AI that is concerned with the development of formal systems that enable the representation, manipulation, and reasoning over information and knowledge encoded in symbolic form. Such systems go beyond mere data storage and are devised to facilitate the connection of information while making implicit knowledge explicit through semantic evaluation [59]. In construction development risk management phases, Okudan et al. [19] found that risk identification, analysis, response, and monitoring are not usually integrated. To address this gap, the authors developed a knowledge-based risk management tool named CBRisk, based on case-based reasoning (CBR). CBRisk is implemented as a web-based tool that supports iterative risk management processes and incorporates an advanced case retrieval mechanism utilizing a comprehensive set of project similarity features expressed through fuzzy linguistic variables. The tool was validated and demonstrated considerable potential for enhancing the effectiveness of risk management practices in real-world construction settings. In [40], a model was developed to enhance the accuracy of AI-driven cost estimation for new projects by leveraging historical data. Among the approaches evaluated, the CBR model demonstrated superior performance due to its adaptability to novel cases, robust handling of structured and complex numerical data, and resilience to incomplete information. Notably, CBR proved capable of retrieving relevant historical cases even when the variable profiles of new projects only partially aligned with those in the case base [40]. The authors also highlighted the drawback of the CBR approach, particularly its challenges in addressing highly complex tasks, as well as the inherent difficulty in transferring case-specific elements between different projects. In another early study, Paredes-Valverde et al. [51] proposed a semantic-based system grounded in symbolic AI, designed to recommend suitable personnel for new software projects by leveraging their experience from previous projects. The approach outlined in this research achieved satisfactory outcomes in both matching similar projects and suggesting appropriate personnel. The primary contribution of this study lies in the integration of semantic reasoning techniques into the personnel assignment process within software development environments, thereby supporting human resource managers in streamlining decision-making and offering the potential for full automation of the task.

4.3.4. Computational Intelligence (CI)

Computational intelligence is a subset of AI according to [60]. Computational intelligence refers to a group of nature-inspired methods used to solve complex real-world problems. These problems are often too difficult for traditional mathematical models due to their complexity, uncertainty, or randomness [61]. Swarm intelligence is an example of computational intelligence techniques. Han et al. [52] presented an optimized resolution for a software project plan using an improved swarm intelligence approach, namely a Max–Min Ant System algorithm. This nature-inspired metaheuristic simulates the foraging behavior of ant colonies and aims to discover the shortest path among tasks. Their method generates feasible solutions to facilitate scheduling tasks based on required skills and available employees, ultimately producing an optimized project schedule. The Flow Direction Algorithm (FDA), another nature-inspired heuristic AI optimization method, is also used in [42] to address the optimization of factors in construction management. The study offers tailored optimal solutions for each factor and balances the trade-offs among them, enabling construction managers to make decisions that align with the project’s original objectives and effectively guide factors toward the desired outcomes.

4.3.5. Hybrid Intelligent System (HIS)

A hybrid intelligent system represents a unified strategy that merges at least one form of intelligent technology with another intelligent method or several approaches to address the shortcomings of individual techniques. The objective of this combination is to improve knowledge representation, enhance reasoning abilities, improve operational efficiency, and accelerate problem-solving for addressing intricate real-world challenges [62].

Fasanghari et al. [50] proposed a computational intelligence approach based on the locally linear neuro-fuzzy (LLNF) model, which is a hybrid intelligent system that integrates fuzzy logic and neural networks to estimate and predict project time and cost metrics using earned value management. Applied to Iranian IT projects, the method demonstrated high accuracy, ease of use, and lower prediction error compared to traditional regression, forecasting, and neural network models. It proved particularly effective for early-stage project status prediction, offering valuable decision-making support for a wide range of project stakeholders. The proposed approach can be adapted for use in other regions with appropriate localization. In a more recent study, Benala et al. [31] proposed the swarm intelligence-based functional link fuzzy neural estimator (SFNE) for software development effort estimation. This hybrid intelligent model integrates interval type-2 fuzzy logic systems, active learning, and particle swarm optimization to improve prediction accuracy and reliability. The findings highlight the robustness of the proposed hybrid intelligent system and its adaptability across diverse datasets, making it a valuable tool for accurate effort estimation in software development projects.

4.4. Summary of Findings for Q2

To answer Question 2, all of the selected literature has been thoroughly reviewed in this study. Natural language processing (NLP), machine Learning (ML), knowledge representation and reasoning (KRR), computational intelligence (CI), and hybrid intelligent systems (HISs) are identified as the five sub-fields under the category of AI. Since several of the remaining studies only refer to AI in general, without specifying particular directions, they are not classified under any sub-fields or algorithms in this study. Therefore, those studies are not included in this explanation section. In addition, all AI algorithms are exhaustively listed under their respective sub-fields (see Table 3). While these techniques offer significant advantages, such as improved prediction, decision support, and process optimization in various industries. However, the limitations of each algorithm are also presented among the reviewed articles. NLP is highly reliant on the language database and query quality [29]. While ML can experience challenges such as overfitting and limited transparency, particularly when employing black-box models [9,43,45], it also requires massive amounts of training data [48]. KRR, including case-based reasoning, may be limited by the quality and thoroughness of available cases [40]. CI often requires considerable computational power and rigorous parameter tuning [28]. While HISs are effective in improving accuracy and flexibility, they also bring additional complexity in system integration [34]. In conclusion, the reviewed literature shows that various sub-fields of AI and their associated algorithms have been widely applied to support project success, and some of them have significant results and contributions to the industry. Nevertheless, since different AI techniques offer distinct advantages and limitations, it is important for future AI practitioners to select the appropriate AI technique by considering the implementation environment.

5. Conclusions

This systematic literature review analyzed a total of 61 journal articles published between 2015 and 2025, aiming to explore the relationship between AI and project success from the existing literature in two key dimensions. This paper offers two key theoretical contributions. Firstly, it identifies how AI techniques support critical factors that drive successful project outcomes. Secondly, it systematically examines which AI sub-fields and algorithms have been applied to project success in recent studies. This review investigates the current extent of AI adoption in PM, and provides a structured overview and analysis of the sub-fields and algorithms in relevant studies. Moreover, through a structured categorization of AI sub-fields and their related algorithms within the context of project success, this study sheds light on a topic that has received limited systematic attention in the current literature.

On the other hand, in terms of managerial contributions, the findings offer valuable insights for project practitioners with limited familiarity with innovative technological tools, suggesting that integrating AI into PM can significantly and positively influence project success. Furthermore, among practitioners with prior familiarity with AI in PM, cognitive capacities and domain knowledge exhibit considerable variation. Consequently, this study provides important insights for project management professionals and relevant stakeholders aiming to harness AI technologies to enhance project outcomes by systematically outlining the strengths and limitations of various AI sub-fields and algorithms in their contributions to project success. In addition, this paper includes typical examples that provide a practical reference framework for project practitioners to assess the applicability of specific AI techniques in enhancing project success across diverse contexts. Lastly, by identifying the most commonly used AI sub-fields and algorithms in project settings, this study enables top management teams and decision-makers to make more informed investments in AI tools tailored to their organizational needs.

Regarding the limitation of the study, this systematic literature review does not focus on a specific industry or a region, which may limit the depth of insights for individual sectors. Since the level of AI maturity can vary significantly across industries and regions, future research could address this gap by concentrating on a particular domain to provide more tailored findings. Additionally, this study focuses solely on project success dimensions without examining the corresponding project management process groups. Future studies may benefit from integrating both success dimensions and AI applications within specific process groups to offer a more comprehensive understanding of AI’s role throughout the project lifecycle. The five process groups defined by the PMBOK Guide—initiating, planning, executing, monitoring and controlling, and closing—can be used as guidance for future project lifecycle investigation [63]. This direction would enable future research to systematically investigate the connections between AI sub-fields and algorithms and various dimensions of project success, as well as project management process groups. By synthesizing these relationships, scholars and practitioners can better understand which AI techniques are most appropriate for enhancing specific success factors or supporting targeted project management processes. Moreover, such insights could guide the selection and implementation of suitable AI tools tailored to different project objectives and phases, thereby promoting more effective and context-sensitive applications of AI in project environments. Additionally, this study focused exclusively on peer-reviewed journal articles, which may have introduced publication bias by excluding valuable insights from gray literature and industry reports. Therefore, it is highly recommended that future studies incorporate such literature to provide a more comprehensive perspective.