AI’s first major economic divide is not emerging between companies that can buy tools and those that cannot. It is emerging between workplaces where managers redesign work and those where managers merely authorize accounts.
The AI adoption gap already looks like a management test: a Brookings research paper that 43% of U.S. workers used AI for their jobs in 2026, compared with 32% among workers in six European countries, with Britain at 36% and Italy at 26%.
On a narrower firm measure, 7% of U.S. firms used AI in production in 2025 versus 4% of EU firms, while Eurostat’s broader enterprise survey found that 13.5% of EU enterprises with at least 10 workers used AI technologies in 2024.
These numbers do not prove that Americans are more creative, or that Europeans distrust technology. They point to a more uncomfortable truth: AI rewards organizations that already know how to reward initiative, spread good practice and hold people accountable.
The Tools Are Available But The Workplaces Are Not
Software spreads faster than management competence. The U.S. Census Bureau found that AI adoption among U.S. businesses hovered between 17% and 20% from December 2025 to May 2026, with 37% of firms employing at least 250 people reporting AI use. The question is what happens after adoption. A chatbot account does not redesign a workflow, challenge a weak supervisor or make a team share better prompts.
That is why the most important AI question inside a firm is rarely “Who has access?” It is “Who is expected to experiment, measure and improve?”
The link between AI and Management Practices is already visible. St. Louis Fed economists report that country-level AI adoption in production processes correlates strongly with World Management Survey scores, and that workers in firms with stronger performance incentives, merit-based promotion and willingness to address poor performance are much more likely to use AI.
The pattern is plain: generative AI use becomes productive when managers make it part of the job, not when employees quietly smuggle it into email, code and analysis.
The Old Productivity Lesson Is Back
Europe has seen this movie before. The European Central Bank notes that U.S. labor productivity per hour grew about 50% from 1995 to 2019, while the euro area grew 28%, and it links part of that productivity gap to America’s stronger ability to create and use digital technologies in production. AI is not repeating the information technology revolution exactly, but it is rhyming with it.
The classic evidence comes from Nicholas Bloom, Raffaella Sadun and John Van Reenen, who found in the American Economic Review that U.S. multinationals in Europe gained more from IT partly because of tougher people management. Their work on technology adoption showed that computers did not magically lift productivity; organizations had to change incentives and routines around them.
Another report reached a similar conclusion in their study of police departments, where organizational change through CompStat-style management made IT investments productive while computers alone did little.
That lesson should make executives nervous. AI vendors sell speed. Boards buy “transformation.” Workers test tools in the cracks between meetings. But the history of IT says the winners are not the earliest purchasers. They are the firms that turn experiments into operating systems.
Europe’s Problem Is An Incentive Problem
The hardest AI bottleneck is managerial courage. One NBER paper argues that Italy’s productivity stagnation stemmed largely from firms’ inability to exploit the ICT revolution, with weak meritocracy in selecting and rewarding managers playing a central role. That diagnosis of European productivity now matters for AI because generative tools magnify the value of judgment, delegation and performance feedback.
The upside is real. Another NBER study of customer support agents found that access to a generative AI assistant increased productivity by 14% on average, with much larger gains for novice and lower-skilled workers.
That kind of AI productivity does not come from replacing everyone with software. It comes from codifying better practice, shortening learning curves and helping less experienced employees perform closer to expert level. The OECD’s work on digital transformation makes the same policy point from another angle: firms need training, organizational capacity and better diffusion mechanisms, not just slogans about innovation.
The management agenda is therefore brutally practical. Reward employees who find useful AI applications. Promote managers who can turn scattered experiments into shared routines. Stop pretending that every department should invent its own AI playbook in isolation. And treat AI governance as a performance system, not a compliance binder.
Leaders looking for a broader, accessible framework can start with a practical AI strategy that treats adoption as an organizational challenge rather than a tool rollout.
The next AI race will be won inside firms before it appears in national statistics. The winners will not simply have better models. They will have better bosses, clearer incentives, faster learning loops and fewer excuses for letting powerful tools sit unused. Europe does not lack talent. It lacks enough organizations built to turn talent into disciplined experimentation. That is the real gap AI is exposing.
