Hold Up! Don’t We Actually Need More PMs, Not Less? 🤔
Greetings from a another Sunday and I hope you are doing well, all The Product Roast subscribers—if your coffee’s ready, let’s get started to discuss today’s topic.
I know many of you, whether early in your product career or feeling passionate about product management in college, are feeling anxious and uncertain about the dawn of AI in our industry. The question of where to start, how to proceed, and whether there’s a place for humans in the product world are more top-of-mind than ever. Today, let’s cut through those worries with a clear-eyed look at the market, and then I’ll share why I believe the future needs more PMs - not less, even if that idea seems unlikely right now.
First, Let’s Take A Look into PM Job Market as October 2025
Entry-level product manager job postings have crashed to a five-year low. AI can generate research summaries, draft specs, and manage backlogs faster than ever, sometimes in minutes, not hours.
But these are some stats from one of the recent researches:
* Global PM job openings actually rose 7.1% last month.
* Entry and mid-level PM roles grew by 8.0%, and remote PM postings jumped 15%. The surge is real even as traditional pathways in junior roles are shrinking. (For detailed data, check out the October 2025 PM jobs report at newsletter.jamesgunaca.com [http://newsletter.jamesgunaca.com/])
What are the Root Causes of these Concerns?
Honestly, I think the fear comes from what our typical workdays used to look like. Most early and junior PMs spent their time buried in operational tasks: writing docs, running & scheduling status meetings, build reviews, backlog clean-ups. When AI swooped in and started doing these tasks such as writing PRDs, designing prototypes, even generating launch visuals in minutes, the default reaction was: “Well, what’s left for me?” It’s a similar experience for developers, who, with tools like Copilot, are now shipping 55% faster. But speed and automation aren’t the end of the PM story. Actually now, they’re just at the beginning of a transition into what really matters.
What the History Tells Us
Okay, now hold on because I’ll quickly dive into history and try to describe what could be next based on what we faced at recent decade.
A useful lens on today’s AI revolution comes not just from tech, but from medicine. In the article “Why AI Isn’t Replacing Radiologists” by The Works in progress Newsletter (see https://www.worksinprogress.news/p/why-ai-isnt-replacing-radiologists [https://www.worksinprogress.news/p/why-ai-isnt-replacing-radiologists]) the prediction that AI would make radiologist jobs obsolete is debunked by real-world data. The logic is simple: as machines become better at interpreting medical images, the need for radiologists actually increases. Why? Jevons paradox, the idea that when technology makes a resource (in this case, expert analysis and interpretation) cheaper and easier to produce, demand for that resource actually grows. When AI made image analysis dramatically faster, hospitals began to order more scans, doctors requested more second opinions, and new areas of expertise (such as advising on AI errors and guiding treatment based on automated results) blossomed. Instead of shrinking, the radiology profession grew. Tasks multiplied. The bottleneck shifted from old manual work to new forms of judgment, oversight, and high-level decision making.
As following image shows; in 2025, demand for human labor is higher than ever & U.S. diagnostic radiology residency positions hit a record 1,208 (up 4% year-over-year) amid all-time-high vacancies, and radiologists earned an average $520,000, 48% more than in 2015.
Let’s learn from other professions to see how automation can expand rather than erase human roles. In “When More Automation Means More Human Workers,” Deena Mousa explores how widespread ATM and spreadsheet adoption didn’t decimate bank tellers and accountants, instead, it grew demand. James Bessen’s research shows that as ATMs cut the cost of basic transactions, banks opened more branches, and teller employment rose 2% annually through the 2000s. Similarly, when spreadsheets slashed accountants’ rote work, clients demanded more complex financial analysis, expanding the profession.
To see full version of the article, go https://newsletter.deenamousa.com/p/when-more-automation-means-more-human [https://newsletter.deenamousa.com/p/when-more-automation-means-more-human].
This phenomenon reflects elasticity of demand: as technology makes a service cheaper or faster, more people use it, and if that increase outpaces efficiency gains, total labor demand grows.
Four conditions must hold:
* Efficiency boost without full substitution
* Lowered effective cost (e.g., faster turnaround of scans or reports)
* Cost reduction drives higher usage
* Demand growth outstrips productivity gains
But automation can also substitute: if AI achieves extreme productivity or completely replaces a role, the substitution effect overwhelms scale, and employment declines as happened with bank tellers after mobile banking adoption.
AI might handle most PM tasks, but if it helps them test more ideas and use data better, human judgment will matter even more. Only when AI fully subsumes every decision-making and strategic function will PM roles truly contract. Just as tellers, accountants, lawyers, and radiologists moved up the value chain into advisory, interpretation, and oversight, PMs must evolve too: from execution and documentation to curation, prioritization, and editorial judgment.
To be discussed in the next section; building is easy; deciding what and why to build is where PMs make their biggest, most irreplaceable impact.
Wait - Why Do We Actually Need MORE PMs?
The same economic forces that drove radiology, banking, and legal services to hire more experts after automation are now at play in product management. AI tools dramatically boost efficiency, accelerating research, spec writing, and backlog prioritization, but they cannot replace the nuanced oversight and strategic vision that PMs provide. As generative models slash time-to-deliver, organizations request more prototypes, experiments, and data analyses, and easier access to insights fuels more product iterations and discovery sprints.
Crucially, the surge in experiments and feature launches powered by AI outpaces the initial productivity gains, creating a “slopware” not “software” (as stated by Kapadi at https://substack.com/@makapadia [https://substack.com/@makapadia]) tsunami of really bad features & products that lack focus on customer value. Majority of app & software market is flooding with nonsense functions or services that try to solve non-existing problems.
In this environment, deciding what to build and, more importantly, what not to build, becomes the most vital and valuable work. Just as radiologists shifted to interpreting AI outputs and guiding treatment, PMs must evolve from execution and documentation to curation, prioritization, and deep judgement.
And more significantly, i believe this whole AI surge is a huge opportunity.
Key Takeaways
Yes, we definitely accept that AI is able to write specs, create visuals, code things and build prototypes etc. It means that “almost anyone” in the world can create things, because now; it is cheap, and getting better day to day.
But in real economy to create real value to touch humanity’s needs, we need more people to do:
* Maximizing value while navigating trade-offs
* Prioritizing work by extracting genuine pain points from noise
* Connecting daily product decisions to long-term business outcomes
* Confidently saying “No” or “Yes” at the right moment
* Exercising real judgment, never acting like a robot
So, building is easier & cheaper than ever, but judgement about what worth to build or proceed with is still expensive and valuable.
This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit theproductroast.substack.com [https://theproductroast.substack.com?utm_medium=podcast&utm_campaign=CTA_1]
