AI Can Replace Jobs. It Can Also Create Employers.

The debate about artificial intelligence and employment often starts inside an established company: work already exists, people are paid to do it, and technology offers a way to do some of it with fewer employees.

The scale of potential change is substantial. The International Labour Organization’s 2025 assessment estimated that one in four workers worldwide held an occupation with some exposure to generative AI. Exposure means that tasks could be affected; it is not a forecast that those jobs will disappear. The ILO concluded that continued demand for human input makes job transformation more likely than full replacement. (ILO, 2025)

That is a real concern. But it leaves out a different starting point.

What happens when someone has a viable business idea but cannot afford the people, tools, or development needed to get it off the ground?

In that situation, AI may help a founder move from an idea to a business—and, eventually, from working alone to employing others.

Both possibilities belong in the discussion.

The Business That Never Gets Started

Myles Munroe captured the cost of unrealized potential in Understanding Your Potential: “Our graveyards are filled with potential that remained potential.” He described the cemetery as a repository of ideas, creative work, and ambitions that never reached the world. (Source: preface)

For an aspiring business owner, that thought raises a practical question: how many worthwhile ideas remain unrealized because their creators cannot afford to take the first step? AI cannot guarantee that an idea will succeed, but it may give more people a way to find out.

That hesitation has a measurable human dimension. In the Global Entrepreneurship Monitor’s 2024 survey, 47% of respondents who saw good opportunities to start a business said fear of failure would stop them, compared with 42% in 2019. These are reported attitudes, not a count of viable businesses lost, but they reveal a gap between recognizing an opportunity and feeling able to pursue it. (GEM 2024/2025 findings)

The decision can also depend on what someone stands to lose. A U.S. study published in 2011 found evidence of “entrepreneurship lock”: employer-provided health insurance could discourage people from leaving employment to start a business. Earlier research using inheritance data also found that access to capital influenced entry into entrepreneurship. These studies examined particular populations and periods, but they show why an unrealized idea cannot simply be dismissed as a lack of ambition. (Fairlie, Kapur, and Gates, 2011; Holtz-Eakin, Joulfaian, and Rosen, 1993)

Behind the decision may be a person protecting a steady paycheck, household savings, or a family’s security. A less expensive way to test an idea could make the first step more manageable without requiring an immediate leap into full-time entrepreneurship. That is one potential human benefit of AI-assisted experimentation—even though it cannot remove financial insecurity or guarantee a successful business.

Research shows how some early work can become less time-consuming. In a 2023 experiment involving 453 college-educated professionals, access to ChatGPT reduced time spent on assigned professional writing tasks by 40%, while independently assessed quality improved by 18%. The experiment tested short writing assignments, not the launch of entire businesses, but it illustrates how AI can reduce the effort required for particular pieces of work. (MIT’s account of the study published in Science)

Consider an independent filmmaker with a concept for an animated series.

Traditionally, producing a convincing pilot could require more money than that creator can raise. Writing, animation, editing, sound, and production all require time and expertise. Without a way to fund that work, the project may stay on paper.

Now imagine that creator using AI-assisted tools to develop a rough pilot, test the concept with an audience, and demonstrate enough interest to attract funding or paying customers.

That early version may have limitations. It may need substantial human refinement before it becomes a commercially successful product. But it gives the creator something concrete to test.

If the project gains traction, its needs can change. An experienced editor may improve the pacing. An animator may bring consistency and character to the work. A performer may deliver something the audience connects with. Someone may be needed to manage partnerships, schedules, or customer relationships.

The initial use of AI could help make those paid roles economically possible.

The important question is what would realistically have happened without it. If the alternative was a funded production hiring people, the employment implications look different from a project that would never have launched.

We should not assume every AI-assisted project replaces an available job. We should not assume every one creates a future job, either.

From Founder to Employer

The same possibility extends beyond creative production.

There is an economic reason to pay attention to new employers. Research by John Haltiwanger, Ron Jarmin, and Javier Miranda found that startups and young firms play a disproportionate role in U.S. job creation. It also found substantial job losses when young firms close. This research predates today’s generative AI tools; it establishes why business formation matters, without establishing that AI will increase it. (Who Creates Jobs? Small vs. Large vs. Young)

A founder might use AI-assisted development to build an early software prototype. A small service business might use it to prepare routine documents and organize information while its owner develops a customer base. An entrepreneur might test a product before committing to the expense of a larger operation.

In each case, AI could lower some of the cost of getting started. It does not remove the need for expertise, judgment, or a product customers actually want.

A possible path looks like this:

An idea becomes testable.
The founder produces something people can evaluate.

Demand becomes visible.
Customers demonstrate whether the offering solves a problem worth paying for.

Revenue supports investment.
The business gains room to improve quality, increase capacity, and address weaknesses.

Human expertise becomes a priority.
The founder hires or contracts with people whose skills help the business deliver and grow.

None of those transitions is automatic. A prototype is not proof of demand. Revenue is not necessarily profit. Growth does not always lead to hiring.

But this path deserves attention alongside the familiar story of an existing employer reducing headcount.

Why Hire If AI Helped You Start?

A business can use AI extensively and still have good reasons to hire people.

Getting an operation started and running it well at scale are different challenges. As customers arrive, expectations increase. Work needs review. Relationships need attention. Decisions carry greater consequences.

A founder who can produce a first version alone may still need someone who can make it reliable, distinctive, or easier to use. The ability to generate more output does not necessarily create the capacity to evaluate it, support it, or take responsibility for it.

Human expertise can become more valuable as the business develops something worth protecting.

That does not require abandoning AI. A growing company may continue using it for repetitive tasks while investing in people for creative direction, technical depth, service, leadership, and accountability.

One study shows what that combination can look like in practice. Research published in 2025 examined 5,172 customer-support agents and found that AI assistance increased issues resolved per hour by 15% on average. Gains varied across workers, and agents retained responsibility for customer conversations. The finding demonstrates a productivity benefit from people working with AI; it does not establish whether employers will use that benefit to expand service or reduce staffing. (Generative AI at Work, Quarterly Journal of Economics)

The business decision becomes: Where would another person make a meaningful difference to what we can deliver?

More Businesses Does Not Automatically Mean More Jobs Overall

There is a limit to this argument.

An AI-assisted business that hires five people has created five roles within that business. That alone tells us little about the wider employment effect.

It may win customers from an established competitor that subsequently reduces its workforce. It may purchase fewer outside services than a comparable business would have purchased before AI. It may remain a profitable solo operation indefinitely.

Lower barriers to entry can also mean more competition. Making a product easier to create does not make customers easier to acquire.

Even business performance gains are uneven. A field experiment giving Kenyan entrepreneurs access to a GPT-4 business adviser found no statistically significant average improvement in revenues and profits. Outcomes differed according to entrepreneurs’ performance before the experiment. Access to advice alone was not enough to produce a reliable benefit across businesses. (The Uneven Impact of Generative Artificial Intelligence on Entrepreneurial Performance)

For those reasons, the possibility of new employers should not be used to dismiss people whose work is being displaced. Future opportunities do not automatically compensate someone for losing a job today.

The narrower claim is still significant: AI may make some businesses feasible that otherwise would not exist. Some of those businesses may become employers.

Whether that produces a net gain in employment is a separate question that requires evidence.

What This Means for Business Owners

For someone building a business, the useful lesson is to connect technology decisions to commercial reality.

Use AI to make an idea testable. Find out whether customers need what you are offering. Understand the cost of delivering it well, including the review and correction that AI-generated work may require.

That review has practical value. A 2023 experiment involving 758 BCG consultants found that AI helped on tasks within its capabilities, but on a task designed to fall outside those capabilities, AI users were 19 percentage points less likely to reach the correct answer. For a business owner, the lesson is to test a tool against the actual work and check the result before treating speed as progress. (Harvard Business School working paper)

Then identify where human expertise would improve the business most.

That might mean bringing in a specialist before launch. It might mean hiring for customer service once demand grows. It might mean using automation to give an existing team more room to take on valuable work.

The right sequence depends on the business. Starting with the goal of minimizing headcount can obscure the capabilities the operation actually needs.

The ADAP Perspective

At ADAP, we believe technology must earn its place in the solution. That applies to a new venture as much as an established organization.

Sometimes its value is helping a team work more effectively. Sometimes it is giving a founder a practical way to begin.

The AI-and-employment debate needs to examine both the jobs that technology may displace and the businesses it may help bring into existence. Neither outcome should be assumed.

For a business owner, that starts with practical decisions: what to build, where technology can help, and when to invest in people. ADAP’s business consulting brings operational experience, business analysis, and hands-on implementation to those decisions. We help you assess workflows, identify useful applications for AI and software, and turn your priorities into a practical plan.

Whether you are testing a new idea or improving an established business, we can help you work through the next step. Talk with ADAP about your business, or call 470-394-8301.