The Rise of AI in Project Management
Across various professional settings, executives and project professionals frequently discuss AI’s potential to revolutionize project portfolio management (PPM). They envision AI crafting business cases, building project plans, and even generating status reports autonomously. These ambitions reflect the tech industry’s promises of efficiency, accuracy, and predictive insights. However, as enticing as these prospects are, the reality might be more nuanced. AI in PPM is a double-edged sword—offering both promise and peril.
AI’s appeal lies in its ability to analyse massive datasets, identify patterns, and automate repetitive tasks. Tools like Microsoft’s Co-pilot are already entering the PPM space, aiding teams by drafting project documents, identifying potential risks, and generating insights. However, many organizations remain cautious. One recurring question arises: “Can we trust an AI-generated project management plan or a critical decision that’s heading to a sponsor or board for approval?” This scepticism is not unwarranted, given the inconsistencies observed in AI-driven outputs and the critical nature of the decisions at stake.
The Limitations of AI in PPM
While AI holds immense potential, it is not without significant limitations. One glaring issue is the reliance on clean, complete, and contextually relevant data. AI tools like Microsoft Co-pilot can be impressive in their ability to analyse and draft content, but their effectiveness hinges on the quality of the data they are fed. Organizations often face challenges with incomplete, outdated, or inconsistent data, which can lead to flawed insights or decisions. For example, an AI might recommend a project timeline adjustment without accounting for external factors like stakeholder concerns or the evolving regulatory landscape.
In my work with a large state government entity, we encountered a unique opportunity to use AI effectively. The organization had years of business case data scattered across various formats—printed documents, PDFs, and Word files. By using machine learning, we digitized and organized this data into a unified system, enabling us to analyse trends, learn from past approvals, and even identify commonalities in successful cases. This process unlocked insights that were previously inaccessible. However, the success of this initiative depended heavily on the quality and preparation of the data. AI didn’t fix bad data—it required us to first establish a strong foundation.
A prominent cautionary tale is Australia’s Robodebt scandal. This initiative relied on automated systems to identify welfare overpayments but failed due to fundamental errors in data interpretation and a lack of human oversight. The fallout underscored a vital lesson: AI cannot replace nuanced human judgment and must be supported by robust governance frameworks.
In the context of PPM, this means AI might identify a project at risk based on historical data patterns but fail to recognize unique circumstances that a seasoned professional would catch. A project management plan generated by AI, while efficient, risks overlooking critical details like team dynamics or strategic priorities if not carefully reviewed by experienced practitioners.
Lessons for PPM Professionals
From these experiences, a few critical lessons emerge for professionals integrating AI into their PPM practices:
First, human oversight is indispensable. AI can analyse data and provide recommendations, but it lacks the judgment required to navigate the complexities of organizational objectives and human dynamics. Professionals must remain the final arbiters of decision-making.
Second, AI should augment, not replace, human expertise. The goal of AI adoption should be to enhance human capabilities. For example, using AI to analyse project risks can free professionals to focus on strategic mitigation plans.
Third, organizations should start small, iterate, and scale responsibly. Begin by deploying AI in low-stakes areas, such as automating progress reports, before scaling its use in decision-critical functions. This approach minimizes risks while allowing teams to build confidence in AI-enabled workflows.
The Path Forward: Responsible AI Adoption in PPM
One truth about AI adoption in PPM stands out: the excitement about what AI can do often overshadows the realities of what it takes to get there. Successful integration requires preparation, collaboration, and a clear understanding of where AI adds value.
Take, for example, a state government client I worked with. They had years’ worth of business case data spread across formats—scanned documents, PDFs, and Word files. By introducing machine learning, we were able to digitize this data, organizing it into a central system that allowed us to uncover trends, insights, and patterns from previous projects. It was a powerful use of AI, but it didn’t happen overnight. The real work was in preparing the data, cleaning it up, and ensuring it was reliable. Without that foundation, even the best AI wouldn’t have delivered useful results.
This is where many organizations falter. They dive into AI without addressing the basics—clean data, solid frameworks, and governance structures. And when AI produces outputs that miss the mark, the technology is often blamed, when in reality, it is the setup that failed.
This leads to a fundamental lesson for anyone considering AI in project management: AI is not here to replace expertise. It is a tool—a sophisticated one, but a tool nonetheless. When used wisely, it can speed up processes, identify patterns humans might miss, and even offer initial insights. But it is not a decision-maker. Whether it is a project management plan drafted by AI or a dashboard presenting risks and forecasts, the final judgment needs to come from experienced professionals who understand the full context.
So, where does this leave us? The path forward lies in balance. AI should augment our work, not take it over. Start small. Experiment with low-risk areas like automating repetitive reporting tasks. Build confidence, refine processes, and scale gradually. And always, always ensure there is a human in the loop—someone to ask the critical questions AI cannot answer and make decisions that align with broader organizational goals.
As we look ahead, the promise of AI in PPM is exciting, but its potential will only be realized through careful, responsible adoption. It is not about chasing trends—it is about using technology as a partner to enhance what we do best: applying creativity, judgment, and experience to deliver successful outcomes