Open Access Vol.
No.
December 2025 https://attractivejournal.
com/index.
php/ajse From Problematic Projects to Smart Solutions: A Literature Review on the Role of AI in Modern Project Management Theresia Avila Bria 1*.
Joko Suparmanto 1.
Onisius Loden 1.
Lodia Semaya Amnifu 1.
Engelbertha N.
Bria Seran 2 1 Politeknik Negeri Kupang.
Indonesia 2 Universitas Katolik Widya Mandira.
Indonesia thessabria12@gmail.
Abstract
ARTICLE INFO
Article history:
Received October 27, 2025 Revised November 30.
Accepted 12 December Despite extensive methodological progress, project failure rates remain persistently high across sectors such as construction, information technology, and public infrastructure.
This study employs a Systematic Literature Review (SLR) based on the PRISMA framework, analyzing 78 peer-reviewed articles published between 2015 and 2025 from databases including Scopus.
Web of Science.
Ie Xplore.
ScienceDirect, and SpringerLink.
The review identifies three primary categories of factors contributing to project failures: .
organizational shortcomings such as weak planning, limited stakeholder engagement, and ineffective risk governance.
external disruptions linked to market volatility, regulatory changes, and environmental instability.
technical and operational deficiencies, including reactive monitoring and resource mismanagement.
Within this context.
Artificial Intelligence (AI) emerges as a transformative enabler in project management.
AI applications are grouped into four domains: early risk detection and prediction, decision support and optimization, real-time monitoring and control through IoT and analytics, and systematic learning from failed projects using knowledge-driven approaches.
While the literature emphasizes AIAos role in achieving project success, this study highlights its corrective and recovery potential for failing projects.
The paper proposes reframing AI not only as a success enabler but as a critical tool for failure prevention and recovery.
Future research should prioritize empirical validation, hybrid humanAeAI decision-making models, and cross-sectoral applications to strengthen AIAos role in building adaptive and resilient project management frameworks.
Keywords: Project Management.
Problematic Projects.
Artificial Intelligence.
Risk Prediction.
Project Recovery Published by CV.
Creative Tugu Pena
ISSN
Website https://attractivejournal.
com/index.
php/ajse This is an open access article under the CC BY SA license https://creativecommons.
org/licenses/by-sa/4.
@ 2025 by Authors INTRODUCTION Project management is widely recognized as essential for organizational success, especially in sectors such as construction.
IT, and public infrastructure.
However, despite advancements in tools and methodologies, failure rates remain high.
Asiedu & Ameyaw report persistent delays, budget overruns, and cancellations across projects .
, while the Project Management Institute estimates that 11% of investments are lost due to poor project performance driven by cost escalation, delays, and scope issues .
These recurring challenges reflect ongoing inefficiencies that undermine stakeholder confidence and strategic outcomes.
Multiple factors contribute to problematic projects.
Internal issuesAisuch as weak planning, limited stakeholder engagement, poor communication, and inadequate risk managementAiare common contributors .
, while external pressures like regulatory uncertainty and market dynamics further exacerbate risks .
Compounding this, traditional monitoring tools often fail to provide timely insights, limiting early corrective action .
As a result, project success rates, particularly in complex environments, have seen little improvement over the past two decades.
At the same time.
Artificial Intelligence (AI) has emerged as a transformative Technologies including machine learning.
NLP, and predictive analytics allow organizations to process large and dynamic project data sets .
In project management.
AI has demonstrated potential to enhance early risk detection, resource optimization, and real-time performance tracking .
Unlike traditional descriptive tools.
AI offers predictive and prescriptive capabilities to support proactive decision-making .
Despite growing interest, existing literature largely focuses on AIAos role in improving project successAihighlighting efficiency and innovation .
Limited research examines how AI can help prevent, detect, or recover from failing projects, even though failure-driven learning and recovery strategies are critical in practice.
This gap is especially relevant in high-risk industries, where construction megaprojects frequently exceed budgets by over 50% .
, and software projects often fail to meet expectations or completion targets .
Emerging AI solutionsAisuch as automated risk analytics and predictive modelingAimay provide a path toward earlier detection and recovery for troubled projects .
Therefore, this paper seeks to: .
review and categorize causes of problematic projects, .
synthesize existing AI-driven approaches addressing these challenges, and .
identify research gaps and future directions for leveraging AI in improving outcomes of troubled projects.
This perspective shifts AI from being seen solely as a success driver to a strategic tool for navigating and correcting failure.
METHOD
The adopted methodology for this study is a Systematic Literature Review.
Therefore, this method will be adhering to the principles of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) framework that guide a systematic process in identifying, screening, and synthesizing studies on a particular subject .
Application of the guidelines also ensures this review has limited bias and enhances the replicability of the findings.
Table 1.
Screening and Selection Process of Literature Review Stage Description Identification Screening Eligibility (Full-tex.
Inclusion (Final Se.
Records retrieved from databases (Scopus.
WoS.
Ie Xplore.
ScienceDirect.
SpringerLin.
After de-duplication, records screened on title and irrelevant studies excluded Full-text articles assessed against inclusion and exclusion criteria Studies meeting all criteria & included in synthesis Number Records 1,024 Figure 1.
PRISMA Flow Diagram of the Review Process This is a PRISMA-diagram of the systematic selection of articles.
In total, 1,024 records from five databases were identified: Scopus.
Web of Science.
Ie Xplore.
ScienceDirect, and SpringerLink.
After removal of duplicates, it was down to 714 unique records that were screened by title and abstract.
Finally, full texts of 142 articles were assessed for eligibility.
78 of these were found to meet the study's criteria for inclusion and synthesized in this review.
RESULT AND DISCUSSION
Problematic Projects: Causes A Synthesis of the Literature Regardless of several decades of research in project management, the number of troubled projects remains consistently high.
Many studies have shown that few projects go awry due to a single cause but as a result of the interaction of organizational vulnerabilities, exogenous shocks, and technical failures.
Drawing on these elements in construction, information technology, and infrastructure identifies common patterns destroying project performance.
Causes in problematic projects are, in this section, categorized as internal organizational, external environmental, technical and operational, and cross-industry.
Factors: Internal Organizational These are only some of the most consistent and avoidable reasons for project failure.
Literature underlines that the core reasons for a troubled project include poor planning, unrealistic scheduling, and insufficient budgeting .
, 15, .
When risk assessments are superficial, managers tend to underestimate complexity and uncertainty, creating plans that cannot withstand real-world challenges.
Besides, ineffective governance structures fail to clearly outline authority to decide, hence fragmented responsibilities with interventions being late when problems crop up.
Another persisting problem is poor communication among stakeholders, which often results in misunderstandings of the project's goals and deliverables.
Scope creep, which is the inclusion of requirements that were not planned, also results directly from poor communication and a lack of alignment among stakeholders.
This is according to .
These internal flaws denote some organizations' structural defects and pinpoint that the reason for poor project governance lies in an unlinked strategic objective.
Without proper internal controls, a project can easily fail even before external or technical factors arise.
External Environmental Factors Projects operate within wider socio-economic and political domains and are therefore susceptible to exogenous shocks.
Empirical research has continuously identified economic instability for instance, inflation and fluctuating foreign exchange rates as a shock to project finance and procurement, as stated by Reddy Anireddy .
Political instability and the frequent alteration of rules and regulations also bring in uncertainty, especially for large infrastructure projects that involve multi-year These are generally beyond the control of the project manager but do have a significant impact on the performance of the project.
Environmental conditions further complicate these external challenges, especially in construction and infrastructure projects.
Natural hazards, such as extreme weather events, earthquakes, or floods, may disrupt the schedule, increase costs, and even compromise safety.
Stakeholder conflicts, especially in public infrastructure projects, are another form of external disruption that involves competing interests of the government, contractors, and local communities in hindering project progress .
,20,21,.
These external factors underscore the importance of resilient project planning and adaptive strategies that allow organizations to adjust to dynamic environments.
Factors: Technical and Operational Most of the technical and operational reasons are closely linked to the tools, methods, and practices applied during project implementation.
Most traditional project monitoring and control systems are more descriptive and less predictive in nature, allowing limited scope for the managers to perceive early warning signals.
This inadequacy becomes all the more pertinent in big and complex projects where delays and cost overruns start slowly and snowball into unmanageable proportions.
Technical complexity and integration problems are among the leading causes related to system failure in IT projects .
Poor design quality, inaccurate site data, and weak resource management commonly lead to frequent rework and cost escalation in the construction industry.
Similarly, safety planning is insufficient, leading to accidents that cascade into harm for workers and disruptions to schedules and budgets.
In both IT and construction, technical shortcomings are exaggerated by the fast speed at which technology changes, with the quick rendering of tools or methodologies obsolete.
Resolution of these challenges requires more advanced systems such as predictive AI-enabled tools capturing early warnings about potential breakdowns in operations.
Cross-Sectoral Patterns While the specific manifestations of project failure vary, several general patterns can be distilled from the literature.
Perhaps the most common involves systematic underestimation of uncertainty.
In most cases, organizations usually assume that effective risk management eliminates uncertainty, leading them to be overconfident in the project schedules and budgets.
Underestimation, in turn, is closely associated with the reactive management style that is, one in which managers act only after a problem becomes critical.
Hald et al.
argue that such reaction, on one hand, reduces the capacity to prevent cascading failures across the dimensions of the project.
Another cross-sectoral pattern is the limited organizational learning from past project failures.
Despite abundant evidence of recurring causes, many organizations fail to institutionalize knowledge gained from previous problematic projects .
, 26, 27, .
This results in repeated mistakes, particularly in industries where high employee turnover disrupts continuity of expertise.
The persistence of these patterns highlights the need for adaptive learning mechanisms.
In this context.
Artificial Intelligence offers opportunities to systematically analyze past project data, identify hidden patterns, and support continuous organizational learning thus reducing the recurrence of failure .
Table 2.
Factors Contributing to Project Failures in Different Domains Category Specific Causes Domains Most Affected IT.
Construction.
Infrastructure Internal Factors Key References Poor planning.
(Haenlein & unrealistic budgeting.
Kaplan, 2.
weak governance.
External Factors Economic volatility.
Infrastructure, (Reddy Anireddy, political/regulatory Public Projects 2.
Technical/Operational Lack of real-time IT, (Adriana N Construction Dugbartey, 2025.
Westenberger et , 2.
Cross-Sectoral Underestimation of risk.
All domains (Shafqat et al.
Patterns reactive problem-solving.
repeated failures due to lack of learning Consequences Understanding the multitasking causes of problematic projects is crucial for any evaluation of the potential contribution of Artificial Intelligence.
Many of the identified causes such as poor planning, lack of real-time monitoring, and adopting reactive approaches are areas where AI can bring substantial added value.
For instance, predictive analytics help to tackle underestimation of risks, machine learning can detect anomalies in project performance, and case-based reasoning enables organizations to learn from past failures.
More about such relationships will be presented in Section 4.
AI Applications in Overcoming Project Challenges Artificial Intelligence has become a catalytic driver in project management, bringing with itself solutions to old, nagging problems causing projects to go haywire.
In contrast to conventional techniques that are heavy on descriptive reporting and subjective human judgment.
AI offers predictive and prescriptive competencies.
This ability enables managers to identify early warnings, make near-optimal decisions, and systematically learn from past failures.
This section classifies AI applications based on their use in four project management domains: early risk detection and prediction, decision support and optimization, monitoring and control, and learning from failed projects.
Early Risk Detection and Prediction It is below the line that perhaps the most appealing area of AI application lies in identifying risks before they snowball into catastrophic project failures.
Machine learning models trained on historical data from projects can predict with high accuracy the probability of potential overruns regarding cost, delays in schedules, and defects in quality.
For example, random forests and gradient boosting are two widely used supervised learning algorithms that have been applied to forecast schedule slippages for both construction and IT projects with a view to enabling managers to take contingency resources well in advance .
It plays a critical role in risk detection using NLP.
These algorithms analyze vast volumes of project documentation that are unstructured, like meeting minutes, progress reports, and stakeholder communications, to reveal warning signals that would have gone unnoticed otherwise.
In other words, the capability of capturing "hidden risks" from textual data extends the reach of project monitoring beyond quantified measures into a more holistic understanding of project health.
Decision Support and Optimization AI has also proved very efficient in managerial decision-making, such as resource allocation and scheduling.
Traditional project scheduling methods like CPM or PERT are usually not able to cope with uncertainty and changes in dynamic environments.
Their ability to dynamically optimize the scheduling by reinforcement learning or genetic algorithms will answer by self-modification of resource allocations in order to reduce delays and cost escalation.
The manager will be able to simulate different scenarios and set the most resilient course of action for the given circumstances .
, 32, 33, .
AI-augmented DSS further strengthens the capability of a project manager in the evaluation of competing constraints regarding different project compromises.
For example, multiobjective optimization models allow managers to balance time, cost, and quality simultaneously.
This proactive capability will make decision-making, which has been so far an after-the-fact process, predictive and adaptive in practice.
Such systems may form the backbone for realigning failing plans in troubled projects and serve to restore performance.
Follow-up and Control Another domain in which AI has been very useful is real-time monitoring and By embedding AI in devices, specifically IoT devices, the project team has the ability to continuously monitor progress, safety conditions, and resource usage in For example.
AI-powered computer vision systems have been installed in construction sites to monitor whether workers are adhering to safety measures and also to detect deviations from planned workflows.
, 36, .
This will reduce dependence on manual inspections and immediate feedback will be provided to project managers.
Big data analytics integrated into AI empowers detection of anomalies in project performance indicators.
As opposed to waiting for monthly reports, managers can have real-time dashboards with things auto-flagged where an anomaly differs from expected This proactive monitoring increases the timeliness and accuracy of interventions by preventing small matters from growing into problematic outcomes of the project.
Finally, the paradigm of AI-driven monitoring at the level of performance indicators will go from lagging to leading.
Learning from Project Failure Probably one of the most underexplored but highly valued uses of AI in this regard is how it can facilitate organizational learning from failed projects.
Traditional project management typically treats failures as isolated events, with limited institutionalization of lessons learned.
AI systematizes this process through techniques like Case-Based Reasoning, where new project challenges are compared to past cases and solutions recommended based on historical outcomes .
, 40, .
This ensures that lessons from failures are not lost but effectively feed into future project strategies.
Similarly.
AI-enabled knowledge management systems identify recurrent patterns of failures across many projects and provide insights beyond the level of a single case For example, clustering algorithms can show hidden correlations between project contexts and outcomes, underpin a deeper understanding of systemic risks.
Such applications move organizations from a reactive to a learning-oriented culture, reducing the likelihood of repeating mistakes Table 3.
Mapping Project Challenges to AI Solutions and Expected Impacts Cause of Problematic AI Application Expected Impact Projects Poor planning & Machine Learning Early identification of unrealistic scheduling .
redictive model.
unrealistic timelines.
Reinforcement Learning scheduling adjustments Ineffective NLP-based analysis of Detection of reports/emails stakeholder misalignment stakeholder alignment Economic & regulatory Predictive analytics.
Simulation of alternative Scenario-based DSS adaptive planning Technical complexity & AI-based optimization & Early detection of integration integration issues anomaly detection automated problemsolving recommendations Lack of real-time IoT AI .
omputer vision.
Continuous performance anomaly detectio.
Limited organizational Case-Based Reasoning.
Systematic reuse of past learning from failures AI-enhanced knowledge prevention of repeated Implications The reviewed literature demonstrates that AI has the potential to address many of the recurring causes of project failure.
By offering predictive insights, adaptive decisionmaking, real-time monitoring, and systematic learning.
AI shifts project management from reactive problem-solving toward proactive control and continuous improvement.
However, these benefits are contingent upon data quality, organizational readiness, and the interpretability of AI models.
The following section (Section .
discusses the strengths, limitations, and gaps of current AI applications in project management.
DISCUSSION