Related post: The Inevitable Shift: AI, Project Management, and the End of Low-Intellect Work
There’s a question every knowledge worker will eventually have to sit with: which parts of your job are you still needed for, and which parts is a machine already doing just as well? For a growing number of professionals, the honest answer is uncomfortable — and it points toward a specific, gradual shift in where you should be investing your time and skill.
A growing group of companies is using experienced professionals to train and improve AI models. Their methods differ, but the objective is largely the same: to make AI capable of performing knowledge work at — or eventually beyond — the level of skilled professionals. Research and practical evaluation of AI-generated work across multiple business disciplines reveal a developing pattern, tested against the dimensions that matter in any job — accuracy, reasoning, completeness, layout and presentation, practical usefulness — and across the full landscape of modern knowledge work: product development, product management, program and project management, vendor management and procurement, HR and recruiting, business intelligence, business strategy, and the entire universe of OKRs, KPIs, dashboards, and performance metrics.
The results aren’t uniform. In some of these domains, AI output is already startlingly strong — polished, structured, and often indistinguishable from what an experienced professional would produce. In others, it consistently falls short in ways that reveal exactly what still requires a human mind. If you pay attention to that gap, it tells you something important about where to place your career bets over the next several years.
Where the Machines Are Already Good Enough
Program, project, and portfolio management — the discipline broadly known as PMO work — is one of the areas where AI has advanced the furthest, the fastest. Much of this work is, at its core, structured coordination: status reports, risk registers, timelines, resource plans, stakeholder updates, dashboards that roll up progress against a plan. These artifacts follow known templates, draw on well-established frameworks, and reward consistency and completeness more than original judgment.
Give AI a goal, a timeline, available resources, established procedures, and measurable outputs, and it can produce a credible plan for execution — often faster and more thoroughly than a person working manually under deadline pressure. A few concrete examples of where this shows up:
- Status reporting and dashboards — pulling scattered updates into a clean, presentation-ready summary of progress, blockers, and next steps, in a fraction of the time it takes a person to assemble the same report by hand.
- Risk and dependency tracking — scanning a project plan to flag likely risks, missed dependencies, and schedule conflicts, then proposing a mitigation plan or governance structure that would take a PM hours to draft from scratch.
- Vendor comparisons and procurement documentation — synthesizing vendor proposals into a structured, side-by-side evaluation against defined criteria, ready to hand to a decision-maker.
None of this means project managers, program managers, or PMO professionals disappear overnight. But it does mean a growing portion of their traditional work — the coordination, documentation, reporting, schedule management, and information consolidation — can now be automated, accelerated, or done by fewer people. The floor for “acceptable output” in this kind of work keeps rising, and AI can now clear that floor with very little friction.
Where Human Judgment Still Wins
The distinction between managing work and deciding what work should exist is becoming increasingly important — and it’s exactly where product work diverges from PMO work. Product decisions require more than processing information; they depend on judgment, curiosity, context, relationships, commercial awareness, and the ability to make difficult calls when the evidence is incomplete or contradictory. A few concrete examples of where AI still consistently struggles:
- Reading between the lines of customer feedback — AI can summarize a stack of customer interviews accurately, but it’s far less reliable at recognizing when a customer’s stated request is masking a different underlying problem, or when a loud, frequent request doesn’t actually reflect a broad market need.
- Owning a strategic trade-off — AI can score a list of features against a prioritization model, generate a roadmap, or simulate a few scenarios, but it can’t independently judge whether that roadmap is the right strategic direction, and it carries no responsibility for challenging the assumptions baked into its own model.
- Navigating organizational politics and unintended consequences — deciding when to kill a popular but low-value idea, redirect investment before consensus has formed, or account for how a decision will land with a skeptical stakeholder requires reading people and context AI simply doesn’t have access to.
Business strategy and product management lean heavily on this kind of tacit, contextual, relational intelligence — and even as AI improves rapidly elsewhere, this remains the area where it continues to fall short.
The Career Implication
Put those two observations together, and a practical strategy starts to emerge: gradually shift your professional development away from the coordination-heavy, plan-execution side of your career — projects, programs, portfolios, PMO — and toward the judgment-heavy, ambiguity-navigating side represented by product work.
For professionals working primarily in project, program, portfolio, or PMO roles, the answer isn’t to abandon that expertise — it’s to build on it and move it closer to product decisions. That might mean shifting from tracking delivery to questioning whether the intended outcome is still valuable; from maintaining a roadmap to helping shape it; from reporting metrics to interpreting what they actually mean; or from managing dependencies to understanding the larger system that creates them. It also means deliberately building capability in product discovery, customer research, market analysis, experimentation, product strategy, and commercial decision-making — before the market forces the transition on you rather than you choosing it.
None of this means walking away from AI. It will keep accelerating research, generating alternatives, analyzing evidence, and challenging assumptions. But there’s a meaningful difference between using AI to strengthen a decision and letting AI make the decision without enough human judgment behind it. The safest career position isn’t the one that competes with AI on speed, documentation, or the ability to produce a polished plan — AI is already very good at that. The stronger position is the one that uses AI effectively while contributing the judgment, context, responsibility, and human understanding the technology still lacks.
The Real Question to Ask Yourself
The question worth asking isn’t simply “can AI do my job?” It’s narrower and more useful: which parts of your work are becoming easier to automate, and which parts require distinctly human capabilities you should keep developing? For many professionals, the answer points away from managing projects as temporary units of work and toward managing products as evolving systems of value.
The transition doesn’t need to happen overnight. But it should begin deliberately. As AI becomes increasingly capable of determining how work can get done, you’ll need to become much better at deciding why the work matters, what should be built, and whether it should be built at all.