The AI revolution promises to accelerate global growth. But the IMF and the OECD now warn of a risk less visible than job losses: countries capable of turning AI into productivity gains could gain a lasting lead over those that merely use it intermittently.
For Tunisia, the challenge is not only to train engineers or to launch platforms. It is to bring this technology into businesses, public services, and occupations, without leaving SMEs and young workers on the sidelines.
The essentials
The IMF estimates that effective adoption could add up to 0.5 percentage points to global annual growth in the medium term; the OECD envisions between 0.4 and 1.2 additional points of annual productivity growth in its member economies. These are potentials, not results already observed.
The World Bank notes that 16.2% of jobs in developing countries could reap substantial productivity gains from AI.
The new global divide is not only technological
On Monday, an employee uses AI to summarize a report. On Tuesday, a small business tests an assistant to respond to its customers. These uses are spreading rapidly. They do not, however, prove that an economy has become more productive. The decisive change occurs when firms alter their ways of working, have reliable data, train their teams, and measure the results.
In a briefing carried on October 22, 2026, Era Dabla-Norris, deputy director of the IMF’s Public Finances Department, stressed developing economies’ ability to adopt and truly diffuse AI. The OECD’s chief economist, Stefano Scarpetta, notes that consumer use still far exceeds integration into the core functions of firms.
The distinction changes the diagnosis: having access to a chatbot does not mean having an industry, a health system, or an administration that is more efficient. Gains depend on the quality of procedures, on training, and on the capacity of institutions to work with new tools.
Substantial potential gains, but not a guaranteed boost to growth
The IMF estimates that effective AI adoption could raise global annual growth by up to 0.5 percentage points in the medium term. This global estimate is not a Tunisian GDP forecast nor a guaranteed 0.5% increase in output for every country.
The OECD projects, for the studied member economies, a potential contribution between 0.4 and 1.2 points to annual productivity growth. The range concerns productivity and covers a set of countries different from those used by the IMF. Adding these estimates would make no sense.
At this stage, empirical work does not yet show a broad acceleration of productivity to the level of these scenarios. Investments in computing power, software, and data centers often precede measurable benefits for user firms.
What the numbers say
The real issue is not choosing between “revolution” and “bubble”. Investments are massive, but the real effects on national productivity remain uneven. Countries capable of training, adapting software, and organizing data will have a better chance of turning technology spending into economic value.
