The AI Economy Still Needs Beginners

At the University of North Carolina at Charlotte, computer science student Jaylen Carroll found an unexpected connection between artificial intelligence (AI) and horticulture. Drawing on his academic interests, he helped develop a digital visitor experience for the university’s Botanical Gardens. The project proved useful enough that the gardens hired him to continue the work.  

It is a modest success story, but perhaps a more revealing one than another billion-dollar AI announcement. A student applied technology to a real problem, discovered its practical value, and earned the opportunity to further develop his skills. This is what workforce readiness should look like.  

North Carolina recently attracted international attention when the G20 Innovation Ministerial convened in Chapel Hill in September. The resulting AI Prosperity Objectives recognized the importance of expanding access to AI training and suggested linking educational opportunities to hiring commitments for workers entering the labor market.  

These objectives alone cannot guarantee jobs. They should, however, help young people develop the knowledge and practical abilities needed to pursue their ideas and create businesses that mean employment. They also raise a question: Who will prepare the next generation to participate in the AI economy?  

As Mark McNeilly, Professor of the Practice at the Kenan-Flagler Business School at the University of North Carolina at Chapel Hill, recently argued in this series, universities need an AI strategy. I would take that argument further. The success of any institutional strategy should ultimately be measured by what students actually learn and whether their education prepares them to contribute beyond campus. The foundation of this strategy should be teaching the guiding principles and specific knowledge of each field, encouraging students to master their disciplines and become experts who can use AI responsibly to make their work more efficient. As Jared Horvath, a cognitive neuroscientist, put it in the National Association of Scholars’s (NAS) podcast, Brains & Bots, “AI is an expert tool.” Thus, schools are obliged to train the next generation of experts.  

That responsibility is urgent because the labor market may increasingly reward experienced workers who already understand how to apply AI within a particular field. These workers can use the technology to complete tasks more quickly, combining responsibilities that once belonged to several employees, including taking on work that previously would have been assigned to junior staff. Employers may therefore have fewer reasons to hire entry-level workers. As the Economist has observed, routine work has traditionally served as an informal apprenticeship through which newcomers develop professional judgment. Meanwhile, some organizations are turning to younger employees to help senior colleagues adopt AI, although technological familiarity does not necessarily imply professional expertise.   

The result is a difficult paradox: AI can make experienced workers more productive while it could reduce the number of opportunities through which beginners can acquire experience.  

The employment prospects image is far more complicated than the familiar extreme predictions of either “mass unemployment” or “unlimited opportunities” suggest. As with many things in technology and life, the truth is somewhere in the middle.  

According to the North Carolina Department of Commerce, approximately 97 percent of businesses already using AI reported no resulting change in employment during the preceding six months. Most reported using AI to supplement workers rather than replace them. That finding describes the current effect of AI on existing employment; it does not show whether employers are changing the way they recruit new workers.  

A separate survey by Western Governors University, drawing on responses from 130 North Carolina employers, found that 35 percent were limiting entry-level hiring because of AI, while 70 percent had changed how they selected candidates. In this context, reduced entry-level hiring means that some employers reported hiring fewer workers for positions typically filled by people at the beginning of their careers. It does not necessarily mean that those employers had eliminated jobs or dismissed current employees. Rather, as mentioned above, AI may allow experienced employees to handle more work, reduce the need for junior support, or change the skills employers expect from new hires.  

These findings measure different populations and different outcomes, so they should not be treated as contradictory. Together, however, they suggest an important distinction: existing employees may retain their positions even as opportunities for beginners become harder to secure. That distinction matters enormously for higher education, because colleges have traditionally prepared students for the first steps of a career in which they gradually acquire professional judgment and specialize in certain roles.  

Consider the legal profession. AI can already produce a reasonably convincing legal memorandum in seconds. But what happens when the cited precedent does not exist or apply, or when a seemingly minor factual distinction changes the entire legal conclusion? An experienced lawyer may recognize the problem. A student who has never learned to reason independently may not.  

Make no mistake: the danger is not that AI can assist with or produce professional work, but that students may mistake the ability to produce such work for the competence required to perform it. That is the gap education still needs to fill.  

This distinction lies at the heart of my research at the NAS. Education should remain more than a process of producing assignments and awarding credentials. A degree should represent knowledge acquired through intellectual effort and struggle—not merely successful completion of a series of tasks.  

AI makes this responsibility more urgent, but it also offers extraordinary opportunities. North Carolina has already taken promising steps to get the AI revolution right. The University of North Carolina System developed a shared course in foundational generative AI literacy and responsible use, making educational resources available across its 17 institutions. This approach can extend access beyond flagship campuses and help students understand the technology they are increasingly likely to encounter professionally.  

Still, AI literacy is a beginning, not an education in itself. Colleges and universities should not signal that AI literacy is the new education. It is only understandable in the field of a profession, by knowing the principles and limits of the field, and being able to use generative AI as a productivity tool instead of a tool to replace the human contribution entirely.  

For example, an engineering student should understand how and why a design works, even if AI generated it. Likewise, a future historian should be able to question AI-generated interpretations of historical events, just as historians have long been expected to critically evaluate books and other sources that have not undergone scholarly review. Simply put, universities should teach students to use AI within their disciplines without sacrificing the independent judgment on which professional competence depends. 

That requires reconsidering assessment. If a chatbot can complete an assignment without the student understanding the subject, professors should ask whether the assignment still measures learning or whether they need new assignments. Students should have opportunities to defend their reasoning and demonstrate how they would approach an unfamiliar problem. Moving essay assignments to oral presentations could help with that.  

Universities should encourage the professional use of AI by teaching students how to evaluate its outcomes, while distinguishing assistance from substitution. This would help manage anxiety around technology by providing the knowledge base needed to build professional confidence for the next generation of experts.  

On this path, North Carolina could go even further by connecting universities with local employers through supervised projects. Students could gain practical experience while helping businesses explore applications they might otherwise lack the expertise to pursue. Just like Jaylen Carroll did. Such partnerships would give employers a role in developing the talent they expect to hire and would give students the practical experience that entry-level positions may no longer provide as readily.  

The state’s growing investment in AI infrastructure makes this educational mission particularly important. For example, Amazon’s $10 billion AI infrastructure project in Richmond County is already under construction, with at least 500 permanent jobs expected. Will North Carolina’s graduates be ready to seize these opportunities?  

The larger opportunity is that if we get the AI revolution right, it may allow smaller businesses to innovate more and graduates to enter professions with capabilities previous generations acquired only after years of practice. We should welcome that possibility without confusing technological assistance with expertise, and without fueling anxiety among young professionals. Equipping graduates with the knowledge and skills is the only (right) way to tackle this anxiety. In this process, students must understand that the best performance is not necessarily the fastest. It will be the outcome that reflects the problem well enough to know its limits.  

North Carolina has an opportunity to demonstrate that technological leadership and serious education can reinforce one another. Its colleges and universities should prepare students to use AI confidently while ensuring that their degrees continue to certify knowledge.  

After all, the purpose of education is to prepare students to accomplish things that neither they nor the machines could achieve alone. 

Lilla Nora Kiss is Senior Fellow for International Affairs & Academic Integrity at the National Association of Scholars.