The employment argument is missing a multiplier
AI Will Create Jobs by Scaling Intelligence
Fewer humans may be needed for each project. But AI can make millions of projects viable that the old economy could not afford to begin.
per project×Vastly more
viable projects
The important question is not only how much labor AI removes from work that already exists. It is how much new work becomes possible when intelligence can be deployed at scale.
When people hear artificial intelligence and jobs, they almost automatically predict the next word: loss.
The assumption is understandable. AI can perform work that once required programmers, analysts, accountants, researchers, designers, and consultants. Some jobs will disappear. Many projects will require fewer human hours. This series does not pretend otherwise.
But that is only one side of the equation. It holds the number of customers and projects constant. It asks how many people will be needed to perform the work we already recognize. It does not ask how much entirely new work becomes viable when customized intelligence can be produced repeatedly.
The project that had no market
Imagine a family-owned restaurant with a scheduling problem that occurs only on Friday evenings during tourist season. The problem depends on that restaurant’s layout, menu, reservation patterns, employee skills, kitchen capacity, weather, nearby events, and the habits of its particular customers.
No software company would build a product for that problem. The market is one customer. The useful life may be eighteen months. The total benefit may be meaningful to the owner but far too small to support a conventional team of consultants and software engineers.
Before AI, the problem sat below the economic resolution of professional intelligence. It remained unsolved—not because nobody could understand it, but because assembling the necessary understanding cost more than the solution was worth.
Now an AI system can help analyze the condition and build the database, forecasting logic, scheduling workflow, employee interface, alerts, documentation, and training materials around that one narrow problem. The system may never be sold to another business. It may exist only for this restaurant, for this problem, during this period of the restaurant’s life. And it can still be worthwhile.
for an industry
A standardized product searches for many similar customers.
for a company
Custom work becomes available to large organizations.
for a department
Intelligence moves closer to the operating condition.
for one problem
Inside one small, unusual business.
Do not hold the market constant
Suppose a consulting engagement once required one thousand human hours and AI reduces the requirement to one hundred. If the number of projects stays fixed, nine hundred hours disappear. But what if the number of viable projects increases one hundredfold?
The economy would use fewer human hours on each engagement while undertaking far more engagements. AI would not merely automate existing professional work. It would place professional intelligence against problems that were previously too small, local, temporary, or peculiar to attract it.
A small manufacturer may need a system for one aging machine no other company still operates. A dental practice may need a cancellation workflow shaped by its school calendar. A plumbing company may need routing software for technicians carrying different equipment across several islands. These problems already contain value. What they lacked was an economical concentration of intelligence.
The jobs move around the machine
AI can perform much of the research, modeling, documentation, and software generation. Humans still have to find the business, earn the owner’s trust, define the problem, secure lawful data access, validate the recommendation, implement the change, train employees, monitor exceptions, accept responsibility, audit the result, and resolve disputes.
- 01Relationship professionals who originate projects and explain the bargain.
- 02Industry specialists who know when a plausible recommendation will fail in a real workplace.
- 03Implementation managers, trainers, validators, auditors, security professionals, and benefit underwriters.
- 04People who carry responsibility for the human consequences of a synthetic recommendation.
This is a crystal-ball scenario, not a labor forecast. The number of new projects is uncertain. Their quality, distribution, and employment intensity are uncertain. Policy and market power will matter. But the mechanism deserves to sit beside displacement in any serious discussion of AI and jobs.
The industrial economy scaled by repeating the same product. The AI economy may scale by regenerating the product around each customer’s exact reality. If that happens, intelligence becomes individually addressable—and the smallest viable unit of professional work becomes the problem itself.
2 thoughts on “AI Will Create Jobs by Scaling Intelligence”