🌏 閱讀繁體中文版
I Was Chatting with Some Recruiter Friends Recently
They said PM positions have changed.
Not fewer openings—just “different requirements.”
I asked: “How so?”
A tech industry recruiter friend said:
“In the past, when I saw ‘proficient in Agile’ on a resume, I’d take a second look. Now there’s too many of those—I don’t even bother.”
“What I look for now is: Has this person handled a mess? Do they have a concrete story about ‘I made a call, and it turned out right/wrong’?”
“There are plenty of PMs who can use tools. Very few who can make decisions.”
That stuck with me for a while.
Big Tech Job Descriptions Are Changing
I looked at recent PM and Program Manager postings from several major companies.
JDs used to look like this:
- Familiar with project management processes
- Proficient in Jira / Asana
- PMP certification a plus
- Able to produce project reports
Now they’re starting to read like this:
Amazon’s Program Manager posting:
“Comfortable making decisions with incomplete information” “Driving projects forward amidst ambiguity”
Meta’s RPM (Rotational Product Manager) program description:
“RPMs are not ‘associate’ PMs—they are full-fledged PMs tackling ambiguous hard problems.”
LinkedIn’s Technical Program Manager requirements:
“Strong situational analysis and decision-making abilities” “Ability to influence without direct authority”
Notice the pattern?
“Able to manage projects” has become table stakes. “Able to make decisions amid chaos” has become the core requirement.
What’s Happening in the Market
According to Project Management Institute data, PM-related positions are projected to grow 33% globally by 2027.
But not all types of PMs are growing.
Let me break down the market demand:
Declining Demand: Execution-Focused PMs
This type of PM mainly does:
- Tracking progress, updating Gantt charts
- Writing meeting notes, organizing documents
- Producing weekly and monthly reports
- Ensuring tasks are completed on time
This work isn’t unimportant, but AI tools can already do 70-80% of it:
- Meeting notes? AI real-time transcription + summary
- Progress tracking? Tools auto-sync
- Weekly reports? AI generates from data automatically
When tools can do most of it, the market value of these skills drops.
It’s not that they’re unnecessary—they’re just “not worth as much anymore.”
Stable Demand: Domain Expert PMs
These PMs have specific industry know-how:
- Healthcare regulatory processes
- Financial industry risk controls
- Manufacturing supply chain management
AI can’t replace this specialized knowledge in the short term.
But long-term? When AI can read all the regulatory documents and learn all the industry SOPs, how long will this advantage last?
Rising Demand: Decision-Making PMs
The value of this type of PM lies in:
- Making judgments with incomplete information
- Coordinating when stakeholders conflict
- Making decisions when direction is unclear
- Persuading people, or saying “no”
Why can’t AI do these things?
Because these require bearing consequences.
AI can give you a pros-and-cons analysis of three options, but it won’t choose for you. AI can help you organize conflicting arguments, but it won’t take responsibility.
“Making a decision”—that always requires a person.
Salary Structures Reflect This Trend
I checked 2024 data from Taiwan’s 104 Job Bank and salary comparison sites:
| PM Type | Annual Salary Range (Taiwan) |
|---|---|
| Entry-level PM (no experience) | NT$400-500K |
| Execution PM (3-5 years) | NT$600-800K |
| Senior Execution PM (5+ years) | NT$800K-1M |
| Decision-Making PM / Program Manager | NT$1.2-2M |
| Senior PM at Foreign Companies | NT$1.5-2.4M |
Notice:
The ceiling for execution PMs is around NT$1M. The ceiling for decision-making PMs can reach NT$2M+.
The gap isn’t seniority—it’s “what capabilities the market will pay for.”
A PM with 10 years of experience who excels at tracking progress might be stuck at NT$900K. A PM with 5 years who can handle complex stakeholder relationships might jump straight to NT$1.5M.
This isn’t about fairness—it’s supply and demand.
There are many people who can track progress. There are few who can make the right decisions amid chaos.
What Role Does AI Play in This Change?
Many people see AI as a threat.
But consider another angle: AI is actually an accelerator.
It accelerates “what was already happening.”
The shift of PM value from “processing information” to “handling people and decisions” wasn’t caused by AI.
Ten years ago, people were already saying: “PMs shouldn’t just be progress trackers.”
It’s just that there was no pressure before, so people could keep doing “comfortable work.”
Now AI is here, and the pressure has increased:
- It handles your busywork → gives you time
- It devalues execution work → gives you pressure
- It changes market demand → forces you to face it
This isn’t necessarily bad.
Many people already knew they should develop toward decision-making—they just never had the motivation.
AI provides a reason.
A Thinking Framework: Where Does Your Time Go?
This isn’t to judge you—it’s for you to see for yourself.
Think about the past week. How was your time roughly distributed?
Category One: Processing Information
- Writing documents, organizing meeting notes
- Updating progress, chasing status
- Producing reports, making presentations
- Replying to emails and messages
Category Two: Handling Uncertainty
- Negotiating, coordinating conflicts
- Making decisions with incomplete information
- Persuading people to accept a plan
- Deciding “what not to do”
- Anticipating risks, handling them early
What’s the rough ratio?
If Category One exceeds 70%, AI might be a threat to you. If Category Two exceeds 50%, AI might be an asset to you.
Scenarios: Things AI Can’t Do
Let me illustrate with specific scenarios.
Scenario One: Client Says “We Need This Feature Next Week”
What AI can do:
- Help you estimate hours
- List dependencies
- Produce an impact assessment report
What AI can’t do:
- Judge “should we accept this”
- Is this client’s request truly urgent, or are they habitually crying wolf?
- If we accept, what happens to the team? To other projects?
- If we push back, how do we say it without damaging the relationship?
These judgments require: understanding the client’s history, the team’s real state, spoken and unspoken political considerations.
Scenario Two: Engineer Says “This Can’t Be Done”
What AI can do:
- Help you search technical documentation
- Find alternative solutions
- Evaluate pros and cons of different approaches
What AI can’t do:
- Judge “is it really impossible, or do they just not want to do it”
- If they don’t want to, what’s the underlying reason?
- Should you push hard, or try a different approach?
- How do you get them willing to try, rather than more resistant?
These judgments require: trust relationships, reading between the lines, knowing when to push and when to yield.
Scenario Three: Boss Says “Why Is Progress So Slow”
What AI can do:
- Help you produce a polished progress report
- List the reasons for delays
- Provide options for acceleration
What AI can’t do:
- Judge “should I tell the truth, or tell them what they want to hear”
- At this moment, how much bad news can the boss handle?
- If I tell the truth, what are the consequences? Can I bear them?
- How do I frame it so the boss supports rather than blames?
These judgments require: political sensitivity, assessment of consequences, and most importantly—willingness to bear responsibility.
Your Choices
After seeing market trends and salary data, you might be thinking:
“So do I have to become a decision-making PM?”
Not necessarily.
This isn’t about right or wrong—it’s about choice.
Choice One: Develop Toward Decision-Making
- High ceiling (NT$1.5-2M+)
- But requires deliberate practice
- Need to proactively seek tasks with “no standard answers”
- Need to bear more risk and responsibility
Choice Two: Move Toward Product
- PM → Product Manager / Product Owner
- From “managing projects” to “managing products”
- Requires more business thinking and user insight
Choice Three: Deepen Domain Expertise
- Become a “domain expert PM”
- Healthcare, finance, supply chain, semiconductor…
- Domain knowledge is a moat, but needs continuous updating
Choice Four: Move Into Management
- From doing projects to leading people
- PMO, department head
- Requires a different skill set
Choice Five: Continue as an Execution PM
This is also a choice.
Not everyone wants to climb up, not everyone is suited for decision-making.
But recognize: this path has a lower ceiling, and competition will only increase.
There’s no right or wrong, but know what you’re choosing.
If You Want to Develop Toward Decision-Making
These things won’t pay off immediately, but will accumulate into your most valuable capabilities.
1. Spend 1 Hour Each Week Thinking “What Could Go Wrong With This Project”
Not tracking progress—thinking about risks.
- Which part is most fragile?
- Who might change their mind?
- What assumptions might be wrong?
Write it down. A month later, look back—how accurate were you?
This trains your “anticipation ability.”
2. Document Your Decisions
Spend 10 minutes each week writing:
- What judgments did I make this week?
- What information did I have? What was missing?
- Why did I choose this way?
- What was the outcome?
Six months later, you’ll have a “decision journal.”
During interviews, this becomes your story bank.
3. Proactively Take On “Gray Area” Tasks
The things no one wants to touch:
- Cross-departmental coordination
- Conflicting stakeholders
- Problems with no clear owner
These are annoying, but completing them grows you the most.
4. Practice Saying “No”
This is the hardest part.
PMs easily become the person who “says yes to everything.”
But a truly valuable PM can judge “what shouldn’t be done.”
Practice refusing one thing each week.
Not refusing for the sake of refusing—it’s practicing evaluating priorities and then taking responsibility for your judgment.
For Those Still On the Fence
If you’ve read this far and still aren’t sure what to do, that’s normal.
I understand your anxiety.
“Learned AI, then what?” “Time saved—will it just get filled with more meetings?” “Where is my value, really?”
These questions have no standard answers.
But one thing is certain:
PMs who don’t think about these questions will be replaced by PMs who do.
Not replaced by AI—replaced by “PMs who use AI + can make decisions.”
If You’re a PM’s Manager
Something to consider:
After your PMs save time, what do you have them do?
If the answer is “take on more projects,” you’ll get more “60-point projects.” If the answer is “spend time thinking clearly,” you’ll get fewer but “90-point projects.”
PM output isn’t “number of projects”—it’s “project success rate.”
AI can help PMs save time, but only you can decide what that time gets used for.
Conclusion
PM work won’t disappear.
But “what PM work looks like” is changing.
AI will continue to decrease the value of “processing information.” But it will continue to increase the value of “handling people and decisions.”
The market will tell us the answer.
Instead of worrying whether AI will replace PMs, look at what kind of PMs the market is looking for.
Then ask yourself: Which kind do I want to become?
You don’t need to have the answer right now.
But starting to think about this question is the first step.
The market will give us time—but not too much.
References
- Compare Salary: How Low Are Taiwan PM Salaries? [Archive]
- 104 Salary Guide: Project Management Personnel [Archive]
- Amazon Jobs: Project, Program, and Product Management [Archive]
- Meta RPM Program Overview [Archive]
- LinkedIn: Program Manager Job Description [Archive]
- IPM: Project Management Careers in 2025 [Archive]