Here’s the thing about the panic over AI taking our jobs: it assumes there’s a fixed amount of work to be done.
Economists call this the "Lump of Labor" fallacy. It’s the idea that if a robot lays a brick, a mason goes hungry. But anyone who has tried to hire a developer in the last five years or waited six months for a specialist appointment knows the truth.
We aren't running out of work. We are drowning in it.
The data tells a clear story. We are currently living in an economy of "unmet demand"—a massive backlog of critical tasks that are simply too expensive for humans to do alone. AI isn't stealing our work; it's finally allowing us to get to the work we've been ignoring.
But—and this is a massive "but"—the way we are integrating this technology is creating a secondary crisis that most executives are ignoring until it’s too late. We are fixing the volume problem by breaking the talent pipeline.
Let me break this down.
The Latent Economy of "Undone Work"
When I was scaling Frndly TV, we didn't stop hiring because we ran out of ideas. We stopped because of capacity. Every business has a roadmap of features, markets, and optimizations they simply can't afford to touch.
This applies globally. According to the World Health Organization (WHO), we are staring down the barrel of a global shortage of 11 million healthcare workers by 2030. That is not a "job displacement" risk; that is a humanitarian crisis that human reproduction rates literally cannot solve.
Look at the legal sector. The Legal Services Corporation (LSC) reports that 92% of civil legal problems for low-income Americans receive no or inadequate legal help. That is a market failure of epic proportions.
This is where the "Latent Economy" kicks in. AI collapses the marginal cost of intelligence, making it economically viable to service these needs. It’s why legal aid organizations are adopting AI at 74%—according to a September 2025 study by Everlaw and NLADA—whereas AI adoption within the legal profession varies by sector, with 37% of e-discovery and personal injury professionals actively using the technology, while overall organizational adoption is cited at approximately 26% to 31%.
When you lower the cost of a service, you don't just get efficiency; you unlock latent demand. 90% of legal aid professionals believe this tech will let them serve clients who previously would have been left to fend for themselves.
So, is AI a job killer? The Information Technology and Innovation Foundation (ITIF) released data in December 2025 showing that AI created 119,900 direct jobs last year while only displacing 12,700.
On paper, we are winning. But if you look closer at who is getting hired, the picture gets ugly.
The Junior Talent Cliff
Here is what keeps me up at night.
While we are busy high-fiving over productivity gains, we are hollowing out the bottom of the workforce. Data from the Stanford Digital Economy Lab shows that employment for the youngest software developers has declined by 20% since its peak in late 2022.
Why? Because of the "Mentorship Tax."
Hiring a junior developer is an investment. It is estimated to consume 20% of a senior engineer's time to mentor a rookie. In the past, we paid that tax because we needed the grunt work done—QA, basic scripts, documentation.
Now, AI does the grunt work.
So, companies are making a logical but dangerous financial decision: stop hiring juniors. Why pay a salary and burn 20% of your staff engineer’s time when a subscription to an LLM costs $20 a month?
The problem is that "senior" engineers don't spring fully formed from the head of Zeus. They are created by suffering through three years of bad code and mentorship. If we cut off the supply of junior roles, where do the seniors of 2030 come from?
The "Workslop" Trap
There is another layer to this. We assume AI makes us faster. But does it?
I've seen this firsthand with marketing and dev teams. Executives believe AI increases productivity by around 25%. But when you actually measure it—as seen in data from 39,000 developers—the realized gain is often closer to 2.1%.
Why the gap? Because of what I call the "Workslop Tax."
Generating code or copy is easy. Verifying it is hard. A study from the Stanford Social Media Lab and BetterUp Labs found that fixing low-quality AI-generated content (known as 'workslop') burdens employees with an average of nearly two hours of additional work per incident. They calculated this "invisible tax" at up to $186 per month per employee.
We are turning our creators into editors. We are turning our engineers into janitors.
Aruna Ranganathan and Xingqi Maggie Ye, researchers at UC Berkeley Haas, recently highlighted a "vicious cycle" where the speed of AI raises output expectations to unsustainable levels. You aren't actually getting more done; you're just generating more noise that requires human verification.
The Cognitive Debt
This leads to the ultimate risk: Cognitive Debt.
Jakub Žegklitz-Bareš, author of "The Economics of Infinite Intelligence," argues that as the marginal cost of intelligence drops to zero, competition shifts to strategic differentiation. But you cannot differentiate if you don't understand the fundamentals.
If a junior lawyer never drafts a simple contract because AI does it, how do they develop the intuition to spot a subtle loophole in a complex M&A deal ten years later? If a junior dev never hunts down a missing semicolon, how do they develop the mental model to debug a distributed system architecture?
We are borrowing against our future expertise to pay for today's efficiency.
What You Need To Do
We can't put the genie back in the bottle. The shortage of 4 million developers (projected by IDC) and the healthcare crisis mean we must use these tools. But we need to stop treating AI as a replacement for entry-level talent and start treating it as a force multiplier for apprenticeship.
Here is your playbook:
- Stop optimizing for headcount reduction. If you cut your junior team, you are eating your seed corn.
- Redefine "Junior" roles. A junior employee's job is no longer "doing the basics." Their job is "learning to audit the AI." Build your training programs around verification and strategy, not rote creation.
- Pay the Mentorship Tax willingly. Recognize that 20% of your senior team's time is the cost of long-term survival. If you don't pay it now, you will pay double for a mediocre senior contractor in five years.
The work is there. The demand is real. But if we let the bottom of the ladder rot away, we won't have anyone left to climb it.