Project Management Pitfalls and the Ambivalent Role of Artificial Intelligence: A Scoping Review of Bad Practices
This scoping review expands the known list of project management bad practices to twenty systemic failures across four key clusters and argues that while artificial intelligence offers potential for improvement, its uncritical adoption risks exacerbating existing issues, necessitating its integration as a complement to rather than a substitute for sound human judgment and governance.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
For decades, the world of building things has relied on a set of rules designed to keep complex efforts from falling apart. Whether constructing a bridge, launching a satellite, or rolling out new software, teams follow a discipline called project management. This field is built on the idea that if you plan carefully, assign the right people, and watch the budget, you will succeed. Yet, despite the invention of sophisticated software and the training of millions of professionals, projects still fail at a stubbornly high rate. They run out of money, miss their deadlines, or deliver a product that nobody wanted. The reasons for these failures are rarely a lack of tools or a lack of knowledge; instead, they often stem from deep-seated human habits. We tend to be overly optimistic about how long things will take, we struggle to say "no" to new requests, and we try to do too many things at once. These patterns are so common that they have been studied for years, forming a catalog of mistakes that seem impossible to escape.
Now, a new question has entered the conversation. As artificial intelligence—computer systems that can learn from data and make predictions—becomes more powerful, many hope it will finally fix these old problems. The idea is that a machine, free from human optimism or fatigue, could calculate the perfect timeline, spot hidden risks, and manage resources without error. But a recent review of the latest research suggests the answer is not so simple. The technology holds real promise, yet it also carries the risk of making our oldest mistakes worse if we are not careful.
A team of researchers from the Technological Park of Holguín and the University of Holguín in Cuba set out to investigate this exact tension. They conducted a scoping review of scientific literature published between 2021 and 2026 across five specific databases: Scopus, Web of Science, DOAJ, OpenAlex, and Google Scholar, looking for studies on why projects fail and how artificial intelligence is being used to fix them. They also examined reports from industry practitioners who deal with these failures every day. Their goal was not to test a single theory, but to map the landscape of twenty specific bad practices that continue to plague the field. They organized these mistakes into four main groups: failures in planning and guessing how long things will take; failures in managing people and the scope of the work; failures in how teams communicate and make decisions; and failures in actually finishing the job.
The researchers found that these twenty bad practices are not new. They are the same patterns described in older studies, proving that the gap between what we know we should do and what we actually do remains wide. One of the most persistent errors is the "planning fallacy," a habit where people convince themselves that their project will be different and better than everyone else's, ignoring historical data that suggests otherwise. Another common issue is the inability to say "no" to changing requirements, which leads to "scope creep," where a project slowly grows beyond its original limits until it collapses under its own weight. The review also highlighted the destructive nature of multitasking, where team members are pulled in too many directions, causing them to burn out and slow down the work.
When the researchers looked at how artificial intelligence interacts with these problems, they found a double-edged sword. On the positive side, the technology shows genuine potential to improve specific tasks. For instance, advanced computer models can analyze vast amounts of historical data to create more accurate estimates of how long a project will take, often outperforming traditional methods. These tools can also track resources to ensure no one is overworked and can automatically flag when a project is drifting off course. In some cases, these systems have been shown to reduce errors in prediction by a significant margin, offering a level of precision that human planners struggle to match.
However, the review warns that simply turning on a computer does not solve the underlying human problems. If the data fed into an artificial intelligence system is flawed, biased, or incomplete, the machine will simply repeat those errors, perhaps even more confidently than a human would. The researchers noted that if a company has a history of underestimating costs, an artificial intelligence trained on that company's past data will learn to underestimate costs as well. Furthermore, there is a risk that teams might become too reliant on these tools, trusting the machine's output without questioning it. This could lead to a situation where bad decisions are made faster and with more authority because they are backed by a computer. The review also pointed out that many artificial intelligence projects themselves fail to reach deployment because they are too complex, solve the wrong problem, or lack clear explanations for how they reached their conclusions.
The core finding of the study is that these failures are fundamentally organizational and behavioral, not just technical. The problems of optimism, poor communication, and the inability to prioritize are rooted in human psychology and company culture. Artificial intelligence is a powerful tool, but it is not a magic wand that can fix a broken culture. The researchers argue that the technology should be used to support human judgment, not replace it. It can handle the heavy lifting of data analysis and pattern recognition, freeing up people to focus on the things machines cannot do: navigating complex political relationships, sensing the morale of a team, and making the difficult ethical choices that often arise in high-stakes situations.
Ultimately, the path forward requires a balanced approach. Organizations that want to succeed must invest in improving the quality of their data and training their people to use these new tools wisely. They must build governance structures that ensure artificial intelligence is transparent and that humans remain in the loop for final decisions. The review concludes that while technology can help us see the pitfalls more clearly, it is up to us to avoid them. The most effective projects will be those that combine the speed and accuracy of artificial intelligence with the wisdom, empathy, and critical thinking of experienced human leaders.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.