Stelian ROMANProject Manager| MicroSafetyCarlingford, New South Wales, Australia
Artificial Intelligence resurgence is taking over the transformation started by Robot Process Automation (RPM) and AI tools move from assisting with isolated tasks to acting as “agents” that (in principle) can plan, execute, and coordinate work.
Can an AI “autopilot” take over for limited periods of time, the project manager job?. In aviation, autopilot works within tightly defined operating conditions, is heavily tested, and always allows human override. By contrast, projects are dynamic social systems: requirements evolve, stakeholders disagree, risk surfaces unpredictably, and success is often measured by outcomes and adoption—not just as within-budget and on-time delivery of the agreed scope.
AI can certainly automate and optimize many activities, but can it govern trade-offs, ethics, and stakeholder trust the way a human project leader does?
In other words, if “autopilot” is possible, it likely depends less on model capability and more on the surrounding system of controls, transparency, and organizational readiness.
Undoubtably, AI is a tool that can and should be used by project managers but, like any tool, it requires knowledge, experience and the case of AI, human intelligence to determine, based on the context, which parts of project management are “automation-friendly,” which require augmented intelligence, and which should remain explicitly human-led because they are fundamentally ethical, political, or trust-based decisions.
With the AI spring and optimistic expectation for a hot summer, a provocative question is emerging in the profession: can project management itself run on AI autopilot—with minimal human direction—while still delivering value, managing risk, and maintaining accountability?
Read also the blog talking about AI as a copilot or a driverless car:
Shenila ShahabuddinPrincipal Consultant| Optimizia INCKarachi, Sind, Pakistan
Thank you @Stelian Roman for the this question.
AI in project management is no longer just a helper. It’s evolving toward “autopilot.” But unlike aviation, projects are messy, dynamic, and human-driven. AI can optimize tasks, flag risks, and even coordinate work, yet trade-offs, ethics, and stakeholder trust still need human judgment.
The real opportunity? Smart collaboration: let AI handle execution and optimization, while humans steer strategy, relationships, and accountability. Autopilot may assist but humans remain indispensable. Saving Changes...
Luis BrancoCEO| Business Insight, Consultores de Gestão, LdªCarcavelos, Lisboa, Portugal
Interesting discussion, both in the post and in the linked article.
The autopilot analogy is useful because it highlights a condition that is often overlooked. Autopilot works in environments that are stable, highly instrumented and governed by strict operating rules. Projects rarely operate under those conditions.
Projects are dynamic social systems. Requirements evolve, stakeholders reinterpret priorities and success often depends on adoption, trust and legitimacy, not only on delivery metrics.
That is why the distinction in the article between AI as copilot and AI as driverless system is so important. AI already performs extremely well as a copilot. It can synthesize information, detect patterns, support risk analysis and expand the analytical capacity of project teams.
But the core of project leadership is not computation. It is responsible judgment. Projects constantly require trade-offs between competing forms of value, and those decisions involve context, ethics and stakeholder trust.
AI can inform these decisions. It cannot legitimately own them.
The real question therefore is not whether projects can run on AI autopilot, but whether organizations are building the governance, transparency and accountability systems required to use AI responsibly.
As analysis becomes increasingly automated, the scarcest capability in projects is not information or calculation. It is accountable human decision-making. In complex systems like projects, analysis can be automated. Responsibility cannot. Saving Changes...
Program Manager| HARPER SRLSanto Domingo / Distrito Nacional, Dominican Republic
Hello Stelian ROMAN , AI can automate many operational aspects of project management, reporting, risk detection, scheduling insights, and data analysis. In that sense, parts of delivery can run in a kind of “assist mode.”
But projects are social systems. Trade-offs, stakeholder trust, and ethical decisions still require human judgment. AI can inform decisions, but accountability and leadership remain human responsibilities. Saving Changes...
Interesting perspective, Stelian Roman. AI can certainly automate planning, reporting, and data-driven insights, but project management also involves human judgment, stakeholder trust, and ethical decision-making. In my view, AI will work best as a copilot, augmenting project managers rather than replacing them on “autopilot.” Saving Changes...
People are going to try. As long as there are cost savings or revenue to be realized, or the belief that either is possible, somebody is going to push the boundaries to see what's really possible. I think the real question isn't "Will AI replace project managers?" but rather "Which project management functions can AI replace and how does the organization need to be changed for this to be possible?"
Slow, organic movement toward increasing AI usage is possible, but if you want change to stick you would have to redesign processes and decision-making to support the use of AI. As PM roles diverge away from administrative tasks and become more about judgment, social credibility, escalation timing, situational interpretation, legitimacy, and managing commitment you are moving into territory where AI does not excel.
Consider different types of PMs and related roles - defense, construction, energy, Sr PMs in other fields, Program Managers... AI can augment some of what they do, but there are a few things that need to be in place for long-term replacement that become more difficult as the roles move away from administrative functions:
- High structure/low ambiguity - Complete, accurate, and timely data - Formalized decision-making - Ownership of AI recommendations - The value of autonomy must exceed the cost of oversight
Leaders like Sam Altman and Sundar Pichai may publicly envision a world where AI can replace them, but they're selling a vision, not a product. This may encourage some to push in that direction. For others, it's vaporware.
I recently saw meme comparing AI to the Thneed from Dr Seuss' book The Lorax. I say this as someone who uses different forms of AI somewhat regularly. While the meme was simple, the comparison fits (albeit a little cynical) - it's a highly marketed, all-purpose tool that consumes natural resources in its production and utilization, produces varying levels of quality output, and may be something that people don't actually need but bought into out of greed or fear (source unknown). Saving Changes...
I got a 404 error for the link, but I would argue that yes, a PM should be able to use AI like an autopilot. As an aircraft systems professional however, that means something very different to me than the typical layperson.
If it’s safe enough to trust with the lives of over 10M people flying daily, why not? The fact is that an aircraft autopilot system is intended as a job aid, not a pilot replacement. First off, it only does a part of the pilot’s job, only under specific situations, and then only with direct supervision. There are many other systems on the airplane which are not under control by the autopilot such as cabin air, radio communication, lighting, etc. In fact the job of the copilot is to manage some of those other functions at times to avoid overloading the pilot. The pilot is still responsible for observing that the system is operating the flight controls as expected. If the pilot needs a break, they don’t just trust the system and walk away. The second-in-command (aka copilot) takes over for the pilot so there is always a human providing oversight.
The system does operate within a well defined set of laws, but aviation is anything but simple. (I did get a chuckle out of the notion that flight sciences, which is literally one component of rocket science, is simple and straightforward.) If something unexpected happens, the pilot immediately steps in and resumes control. Although the autopilot on some modern planes can fly an entire route, takeoff through landing, it is only performing the tasks under routine conditions with multiple types of warnings when things are no longer routine. The ability of the system to operate as expected within defined limits is exhaustively evaluated for different types of risks, tested extensively first via simulation, and then in an operational environment with extremely skilled pilots before it is approved for use. The risks are very well understood and every effort is taken to mitigate them. Sometimes it flies more precisely than a human because it can constantly receive and react to incoming data as opposed to checking and adjusting at less frequent intervals such as when flying in crowded airspace.
I would not agree that AI should be used as an autonomous PM in charge of all aspects of a project. Taking into consideration for what specific functions an autopilot controls, how, and under what set of conditions, I would say yes. It should be able to handle certain well defined repetitive tasks under normal conditions where it has been proven to operate as expected, and while there is someone constantly checking to see if anything has gone wrong and who is fully qualified to take over control. Saving Changes...
Community Champion
