AI: Uzbekistan risks not robots but opportunity gap

Tashkent, Uzbekistan (UzDaily.com) — Uzbekistan should not fear that artificial intelligence will eliminate jobs. It should fear that AI will create two countries within one — a country where technology accelerates growth, and one where it simply does not exist. That is the central conclusion of a report by the Institute for Macroeconomic and Regional Studies (IMRS) on AI risks for Uzbekistan.

According to the report's authors, the primary risk is not mass displacement of employment but an accelerating gap in productivity, income and opportunity between those who gain access to AI tools and those who remain outside that access. Accordingly, IMRS argues that policy must be not merely "about AI" but about inclusive AI adoption.

International data supports this logic. The IMF records that nearly 40 percent of jobs worldwide are exposed to the influence of artificial intelligence — in advanced economies that figure reaches 60 percent, while in low-income countries it stands at just 26 percent. It might seem that Uzbekistan has little to worry about. Yet the fund warns that AI widens inequality not only in labor income but in capital as well. Those with skills and infrastructure see their incomes grow faster. For the rest, the gap deepens.

The OECD adds an important nuance. Generative AI is already being used by 31 percent of small and medium-sized enterprises worldwide, and 65 percent of those report productivity gains. At the same time, 83 percent of companies so far see no impact on their overall headcount. In other words, mass layoffs are not happening — but those that have adopted AI work more efficiently and earn more. For Uzbekistan, as IMRS notes, a paradoxical conclusion follows: the priority should be not combating unemployment but accelerating the spread of AI across businesses — above all among small and medium-sized enterprises.

The World Economic Forum spells out the scale of the approaching transformation: 86 percent of employers expect AI to substantially change their businesses by 2030. For every 100 workers, 59 will require retraining. And 11 of those risk never receiving it. This gap — between those who will retrain and those who will be left behind — is, in IMRS's view, the key fault line of future social stratification.

Without targeted policy, AI could simultaneously accelerate growth and deepen domestic divisions along the lines of skills, geography, gender and business size.

A Regional Leader With an Uneven Start

Against the backdrop of its neighbors, Uzbekistan appears to be in a strong position. According to UNDP data, in the Government AI Readiness Index 2024, the country scored 53.45 points, ranked 70th in the world and took first place among Central Asian states. The country has already entered what IMRS calls a "window of opportunity" — a moment when the right policy decisions can define the trajectory of development for decades to come.

Yet starting conditions within the country are strikingly uneven. Only 66 percent of the population is covered by 4G networks, with full 4G connectivity concentrated primarily in large cities, which are home to around 43 percent of citizens. In rural areas, internet use and digital services remain significantly lower. Women account for just 11 percent of students enrolled in IT programs — a disparity that, in a climate where digital competencies are becoming the primary driver of career advancement, risks becoming entrenched and growing wider.

The position of small and medium-sized businesses is a source of particular concern. SMEs account for roughly 92 percent of all enterprises in Uzbekistan, provide approximately 75 percent of employment and generate more than half of the country's GDP. They are the backbone of the economy — and simultaneously its most vulnerable link in the race to adopt AI, as businesses in this segment face greater constraints on access to capital, computing resources and qualified specialists.

What IMRS Proposes

The report sets out four concrete measures designed to make AI adoption inclusive.

The first is to provide every mahalla — a local community unit — with access to a unified AI assistant capable of advising citizens on public services. Implementation would involve integrating existing government services into a single platform and adapting the interface for users with limited digital literacy. Similar solutions are already operating in China and South Korea, where AI assistants in municipal services have significantly reduced the administrative burden.

The second measure is the creation of AI Sandbox Zones — dedicated regulatory sandboxes for regions where innovative solutions can be rapidly tested and deployed under a special legal framework. The British experience is instructive here: the Financial Conduct Authority's regulatory sandbox has allowed more than 700 companies to test their products without the risk of facing the full weight of regulatory compliance.

The third measure targets the gender and age gap directly: an AI internship program for women and young people, with employer subsidies, a mandatory minimum training period of at least six months and co-financing of wages tied to employee retention outcomes. Singapore's experience, cited by IMRS, demonstrates that income support during training substantially raises participation and lowers barriers to entry into new professions.

The fourth measure is a transitional insurance mechanism for workers whose employment or income may be affected by AI adoption. It would be funded through a dedicated levy on large employers — ranging from one to two percent of the total wage bill — with rebates for companies that retrain at least 70 percent of affected staff. A comparable mechanism exists in the United Kingdom in the form of the apprenticeship levy, set at around 0.5 percent of payroll. In Singapore, support programs provide income replacement of up to 70 to 90 percent during the retraining period.

The common thread running through all four measures is inclusivity. Uzbekistan ranks first in Central Asia for AI readiness. The question is no longer whether the country will manage to enter the technology race. The question is who that race will be run for.