AI adoption threatens expertise pipeline, warns NATO researcher

Nolan Lovett of the NATO Special Operations University published research in the journal Human Resource Development Review arguing that AI adoption is triggering a hidden expertise crisis. His core argument centers on how companies capture the full efficiency gains from replacing entry-level positions with AI, while the cost of eroding expertise is spread across all organizations that recruit from the same talent pool. Because no single employer bears the full cost of expertise erosion, none has adequate incentive to maintain the training environments where junior workers gradually build deep domain knowledge through mistakes, feedback, and independent cognitive work.

Lovett identifies two mechanisms through which AI disrupts expertise renewal. First, AI systems directly eliminate entry-level positions, removing the training ground where beginners once learned. Second, even when junior positions survive, AI assistance lets workers quickly reach productivity levels that historically required years to attain—without the cognitive effort that actually builds deep expertise. This creates a "validation tether" problem: the ability to oversee AI systems effectively depends on exactly the kind of deep domain knowledge that AI use is eroding away. Spotting domain-specific errors in plausible-looking AI output requires more than catching obvious contradictions; surface-level checks aren't enough. Cognitive habits worsen the problem—people who routinely treat AI answers as reliable lose the reflex to question them, and the training environments where junior workers once learned to challenge authority and test claims are vanishing.

The damage timeline is long and hidden. Experienced professionals on the job market today were trained 5 to 20 years ago. Lovett calls this the "Human Reserve Paradox": organizations need deep expertise in reserve for validation, crisis management, and situations that overwhelm AI systems, but no single organization has enough economic incentive to maintain that reserve. Effects from entry-level positions cut starting in 2023 may not fully show up until somewhere between 2030 and 2045. Even workers who survive the pipeline will have shallower expertise because they spent their careers orchestrating AI rather than doing independent cognitive work.

Not all professions face equal risk. Software engineering, financial analysis, and legal research show high task substitutability, relatively light regulation, and strong modularity, putting them in the highest vulnerability category. Medicine and engineering get some protection from stricter regulatory requirements and stronger professional associations. Research on cognitive effects backs Lovett's concerns: an MIT study using EEG measurements showed that even brief AI use weakened neural connectivity, and over 80 percent of participants struggled to recall content from their own AI-assisted writing. An Anthropic study with software developers found that participants with AI access scored 17 percent worse on knowledge tests. A Swiss study of 666 participants found a strong negative link between AI use and critical thinking, with the effect most pronounced among 17- to 25-year-olds. Among Chinese students, homework grades improved by 18 percent but exam performance dropped by up to 24 percent. What matters is not just how much people use AI but how they use it: those who treat AI as a substitute for thinking lose cognitive abilities fastest, while targeted use as an explanatory tool significantly reduces negative effects.

Lovett argues the tragedy isn't inevitable. He proposes that professionals need AI-free learning environments, phased AI introduction, and a baseline of human performance that should come before AI gets involved. Professional associations should test domain competence through certifications alongside AI skills, and policymakers should make training and education more attractive. Bans or restrictions on AI use aren't among his proposals. Employment data paint a mixed picture: a summer 2025 study showed employment declines in AI-affected occupations, especially among young workers, while more experienced workers in the same fields saw employment hold steady or grow. A Federal Reserve Board study found that programming job growth has nearly halved since ChatGPT launched, though a clear causal link cannot yet be proven. Other factors like tighter monetary policy and pandemic-era tech overhiring corrections could also play a role.