AI in Project Management: Five Takeaways from NSPM on Leading Responsibly
Erwin Limon • Manila • September 2026
Artificial intelligence (AI) is changing how project work gets done. But its value depends on more than what the technology can do—it depends on how project managers guide its use, evaluate its outputs, and remain accountable for its consequences.
That message connects attendee reflections from the National Symposium for Project Management (NSPM), organized by the PMI Philippines Chapter at Shangri-La Cebu on August 14–15, 2026. Under the theme “From Profession to Purpose—Building the Nation Through Project Success,” the symposium provided a setting to examine AI through the lens of leadership and project outcomes.
Drawing on contributions from four attendees, these five takeaways explore responsible AI adoption—from governance and human oversight to generative AI, agentic workflows, and coordinated AI systems. They represent a synthesis of attendee reflections, not an official symposium position.
1. Lead with AI, Rather Than Compete with It
One reflection highlights a shift in mindset: project managers should prepare to lead with AI rather than compete against it. As AI becomes embedded in everyday work, the role increasingly involves coordinating people and AI capabilities within a shared workflow..
This expands the leadership challenge. Project managers must determine where AI adds value, how responsibilities should change, and whether the resulting system remains aligned with the project’s purpose.
The question is therefore not simply which tasks AI can perform. It is how work can be redesigned to combine technological capabilities with human strengths while preserving accountability.
Recommended action: Identify one workflow that could benefit from AI assistance. Define the intended outcome, the responsibilities that remain human, and the person accountable for reviewing results before selecting a tool.
2. Build Governance and Ethics into the Project Lifecycle
Attendee notes from the ethical project management discussion emphasize a concern: organizations are learning to use AI faster than they are learning to govern it.
An important distinction follows. Risk does not arise only when AI fails; it can also arise when AI successfully pursues the wrong objective. Optimizing a narrow target without considering wider consequences may undermine the project’s actual purpose.
The reflections position transparency, fairness, privacy, security, accountability, reliability, human centricity, and sustainability as project requirements—not considerations to add at deployment.
They also frame trust as a project outcome. Technical delivery alone is insufficient if employees or stakeholders lose confidence in the resulting system.
Recommended action: Add an AI checkpoint to project planning. Ask whether AI is appropriate, whether the proposed data should be used, who benefits or bears risk, which decisions may be delegated, and what happens when the system is wrong.
3. Make Human Oversight Meaningful
Keeping a person “in the loop” is not enough if that person cannot understand, question, or stop what the AI is doing.
The attendee takeaways identify four conditions for meaningful oversight: competence to understand, permission to question, authority to override, and a clear escalation pathway.
This connects with another recurring theme: calibrated trust. Teams should avoid both accepting AI outputs uncritically and dismissing useful assistance without evaluation. Trust should reflect demonstrated performance, the task involved, and the consequences of error.
For generative AI, convincing language is not proof of accuracy. Review remains necessary even when an answer appears polished and complete.
Recommended action: Assign a reviewer with both the capability and authority to challenge AI outputs. For generated content, check whether it is traceable to sources, complete, appropriately explicit about uncertainty, and actionable.
4. Explore Practical AI Workflows, Not Just Impressive Demonstrations
One attendee highlighted Peter Au’s demonstration of an AI bot connected to Monday.com. Through WhatsApp messages, the bot could update a task and provide insights into its status.
The example illustrates how conversational interaction can become part of project work—not merely a separate activity for generating text. It also raises practical questions about how an assistant interprets requests and translates them into actions.
For project managers, a demonstration is a starting point for evaluation, not evidence that a capability is ready for every workflow. The relevant test is whether it performs the intended task reliably within appropriate boundaries.
Recommended action: Test an AI assistant against a small set of realistic scenarios. Verify that it identifies the correct task, interprets instructions accurately, and reports status consistently with the project record. Require approval before consequential changes.
5. Increase Autonomy Only as Fast as Controls Mature
The reflections describe a progression from prompting to agentic AI, which performs delegated steps, and orchestration, which coordinates multiple agents. Each level requires stronger review practices.
The notes recount an AI note-taker interpreting a tabled discussion item as resolved, causing it to disappear from future agendas. Manual review caught the error—showing how a language interpretation mistake can affect subsequent work.
Multiple agents introduce another challenge: conflicting recommendations. A scheduling recommendation may clash with a funding constraint unless priorities and decision rights are clear.
Recommended action: Establish a delegate–review–correct cycle before expanding autonomy. Define dependencies, approval points, and escalation rules. If the team cannot reliably audit its current workflow, keep the next iteration simpler.
Next Step: Put AI Learning into Practice
PMI Philippines is organizing Happy Hour x Happy Learning: GenAI Hackathon Edition, a follow-on to NSPM.
The event takes place on October 2, 2026, from 12 PM to 6 PM, in BGC, Taguig, for PMI PH members, with a focus on generative AI for project management.
Bring a practical project challenge—and consider how clear objectives, disciplined review, and meaningful human oversight could shape a responsible solution.
Contributors
Thank you to the following NSPM attendees for sharing the reflections that informed this article: Ada Rhea Cura, Maria Juvy Estrada, Erwin Limon, and Albert Santos.
xx
Comments