In a report last year, the United Nations Conference on Trade and Development (UNCTAD) warned that AI’s gains will flow disproportionately to those with strong digital foundations and easy access to compute, data, and talent unless governments intervene. We can see the same pattern in a company, the teams with better tools and better training pull ahead, while everyone else falls behind. If this gap is not addressed, the overall productivity of the organization drops and operational risk increases. If it closes the gap, the organization creates the conditions for AI-driven gains to compound across departments.
UNCTAD projects that the AI market could reach $4.8 trillion by 2033, around the size of Germany’s economy. UNCTAD also estimates that AI could affect a large share of workers, including those in cognitive sectors that many once assumed were safe from automation. The real question for every organization is not whether AI will shape their business, it is whether they can build an institution where most employees can use AI safely and effectively.
Most companies now use AI in their daily operations, but productivity growth is less than expected due to a simple reason. A handful of power users learn the tools and build pilots that impress leadership. But then momentum stalls as most workers cannot access approved tools reliably, do not know what they can share, or what they should avoid. Ultimately employees distrust the tools and abandon them, or they trust them too much and create errors that force rework. Leadership often blames culture when adoption plateaus, but often the real problem is that the company never built the operating foundations for scale.
According to UNCTAD, inclusive adoption rests on three pillars: infrastructure, data, and skills. Infrastructure comes first because it determines whether GenAI feels like friction or flow. Usage collapses if response times lag, endpoints get blocked, logins fail, or access differs by location and device. Uneven internal conditions can create “AI deserts” even in a U.S.-based enterprise. Inclusive infrastructure means the company designs GenAI access so it is consistently available, and secure enough for everyday work.
Data is the second foundation. GenAI only creates value when it helps employees find the right information and take the right next step. UNCTAD says data is the lifeblood of AI systems and highlights how limited access to high-quality, relevant data blocks progress. Inside companies, the problem is rarely a lack of data. It is a lack of trusted, ready-to-use data that can be used safely.
Skills are the third foundation, and they are often the most neglected. Training determines whether AI becomes a daily assistant or a novelty. UNCTAD argues that inclusive outcomes require a worker-centered approach that builds digital literacy, upskills people whose jobs can be augmented, reskills those whose tasks change, and involves workers in implementation. Productivity improves when employees learn how to collaborate with AI reliably and safely, not when they collect prompt tricks.
Workplace research points in the same direction. A recent study from the London School of Economics and Political Science’s The Inclusion Initiative, conducted with Protiviti, reports that AI users save significant time each week, yet most employees surveyed had not received any AI training in the prior year. The same research argues that training, not age, drives adoption and productivity. The lesson is an organization can’t tool its way into productivity without workforce enablement.
Effective upskilling can’t be achieved with a one-time session on prompt engineering; upskilling has to be tied to real workflows. Customer support teams need to draft and personalize responses; analysts need to synthesize, validate figures, and avoid reasoning errors. Sales teams need messaging that stays consistent with positioning and legal guidance, and developers need guardrails for testing and security. Inclusive AI means equipping every role to use AI appropriately, not just enabling early adopters.
When companies connect infrastructure, data, and skills, they move from pilots to measurable gains. Reliable access keeps experimentation inside approved environments. Trusted data grounds output in reality and reduces rework. Training turns experimentation into habit and spreads benefits across the workforce.