Picture: Eoin Ryan
Kathmandu’s traffic police last year deployed AI-powered cameras to spot traffic violations. Bureaucrats speak of digitisation; banks experiment with chatbots; ride-hailing apps dabble in recommendation engines. Nepal is not preparing to build artificial general intelligence (AGI), nor does it need to. The race that matters is not the one between Silicon Valley and Shenzhen but the more consequential race to embed artificial intelligence across the everyday economy. In this competition the country has a fighting chance.
AI is not merely another software upgrade. Rather it is a general-purpose technology with disruptive potential on par with steam, electricity and the microprocessor. The countries that harnessed those earlier waves of innovation were not always the first to invent. They were the first to diffuse. America overtook Britain as an industrial power in the 19th century not through superior science but through superior engineers who applied technology with practical rigour.
Nepal can act on this principle. The Himalayan republic need not train a generation of AI researchers to benefit from the technology. It needs electricians in Pokhara using predictive diagnostics, accountants in Birgunj automating audits, teachers in Nepalgunj deploying personalised learning tools. AI’s productivity gains come, not from breakthroughs in labs, from transformations on shop floors and office desks.
The first priority is pedagogy. Business leaders must learn how AI can be used to rewire operations. If factory layouts needed redesigning to exploit electric dynamos, business processes must now evolve to centre on AI models. Delegating it to the IT department is a recipe for half-measures.
The second task is more urgent: preparing the workforce. AI has a steep learning curve but the bulk of Nepal’s 2030 labour force is maybe employed by now. Waiting for the school curriculum to catch up will miss the moment. Targeted, modular training is required, flexible enough for a textile worker in Itahari, but rigorous enough to meet global standards. Public universities have neither the reach nor the responsiveness to meet this need alone.
Here the private sector and international donors can intervene. Tech firms could emulate Cisco’s Networking Academy or IBM’s SkillsBuild, tailored for domestic contexts. A credible certification regime, one that employers trust, will be essential. Much of the market is flooded with flashy but vacuous courses. A government-backed kitemark for quality could cut through the dysfunction.
Mass skilling would also temper public unease. AI, like most invisible technologies, invites suspicion. Fear of job displacement and algorithmic bias is real and often amplified by the media with a taste for alarm. If citizens experience AI as empowerment, not displacement, the mood will change. Here India provides a template. Its “India Stack”, a suite of public digital infrastructure including Aadhaar and UPI, brought millions into the digital economy and helped build trust in new technologies. Nepal’s fledgling national ID system and digital payments network could play a similar role, if designed with clarity and ambition.
The state itself must do more than watch. With public spending accounting for a quarter of GDP, government operations remain a vast and underutilised sandbox for AI. Procurement could be streamlined through automated fraud detection; land records could be digitised and searched with natural language queries; healthcare diagnostics could be triaged through AI tools in rural clinics. Every ministry has a data problem. Every data problem is an AI opportunity.
The final hurdle is regulatory. Much of the world is racing to build rules before building capacity. The European Union’s sweeping AI Act may reassure its voters but has spooked many firms with its heavy compliance burdens. Nepal should resist the temptation to mimic such models wholesale. The objective should be to encourage adoption and not inhibit it. A lighter, risk-based regulatory approach, focused on transparency and accountability rather than pre-emptive bans, would be more suited to its developmental needs.
There is one realm where regulation cannot be light-touch: the frontier. Although the country is unlikely to develop cutting-edge AI models, it will import them. Ensuring that these are evaluated for safety—particularly in cybersecurity, bioinformatics, misinformation—is vital to maintaining public trust and national security.
Nepal has spent years debating how to accelerate economic growth. Here is the closest thing to a cheat code it may receive. AI will not fix broken roads or clean Kathmandu’s air but it could boost productivity across agriculture, services and manufacturing. It could help track subsidies; match jobseekers with employers; and teach in languages most Nepalis actually speak. Every government since the republic’s founding has pledged to leapfrog. This is the platform from which to try.
The AI age will not wait for laggards. One country will reach AGI first. But many others will thrive by becoming fast followers. Whether the country joins them depends less on the coders of Pulchowk and more on the decisions made in Singha Durbar, local classrooms and factory floors. The race is on. ■