A robust foundation must precede sweeping regulations
KATHMANDU, FEBRUARY 21
Nepal can take inspiration from initiatives like the Singapore–Rwanda AI Playbook, forging strategic, cross-country partnerships that pool resources and expertise.
Is AI the 'Amrita', the nectar of immortality, or must it first pass through the churning of the cosmic ocean-unleashing both divine wisdom and unforeseen chaos before it becomes a true elixir for humanity?
AI, like the fabled 'Amrita', holds the potential to be a nectar of economic and social prosperity, granting Nepal unprecedented opportunities in governance, healthcare, education and business. However, just as the cosmic ocean had to be churned before yielding its divine elixir, AI must also undergo rigorous scrutiny, infrastructure strengthening and regulatory clarity before it can be harnessed for the collective good. The National Artificial Intelligence Policy 2081 aspires to regulate and facilitate AI in Nepal, but does it provide a concrete roadmap to ensure AI serves the country rather than overwhelming it?
The policy outlines broad objectives but fails to address how they will be implemented. It lacks a clear action plan detailing the responsible agencies, funding mechanisms and measurable outcomes. A well-crafted policy must go beyond abstract goals and specify how Nepal will build an AI ecosystem that aligns with its economic and technological realities. Otherwise, Nepal risks falling into the trap of ambitious policies that remain only on paper.
Nepal's AI policy draft emphasizes university collaboration, yet research and development remain weak due to inadequate funding and institutional support. Policymakers highlight academic partnerships, but the reality is starkly different, as seen in the struggles of the National Innovation Centre. India's IITs have propelled a startup boom with over 100 unicorns, showcasing the power of strong educational ties. Meanwhile, Nepal allocates less than 1% of GDP to R&D, leaving institutions like NAST underfunded. Without a clear financing strategy for AI labs, research hubs, and curriculum integration, Nepal will remain reliant on foreign solutions, missing the chance to nurture domestic talent and innovation.
Beyond research, Nepal's broader digital ecosystem remains underdeveloped. The policy assumes AI can flourish without addressing core digital weaknesses such as poor internet connectivity, frequent power outages and inadequate access to computing power. As of 2023, Nepal's internet penetration stood at approximately 50 per cent, with rural areas facing even lower access rates; while mobile phone penetration is relatively high, reliable broadband coverage is still limited to a few urban pockets.
In addition, digital literacy remains a challenge-many citizens struggle with basic online services, raising the question of how they will navigate the far more complex "language" of AI. The AI ecosystem also demands stable and sufficient energy supplies. Without investing in AI-ready infrastructure-such as a robust literacy plan, improved digital connectivity and uninterrupted power, AI initiatives will struggle to scale beyond a handful of tech hubs.
The current draft lacks clarity whether to build our own foundational AI model or leverage the open source solutions. This is critical for Nepal, like many other low-income countries. The introduction of Deepseek, a foundational large language model (LLM), developed in China, showed that the cost of training LLMs can go down significantly.
However, in any case, the need for robust data pipelines, high-powered computational infrastructure and skilled resources is still relevant. A strategic blend of public-private partnerships, government-backed investments and incentives to attract foreign investors could help Nepal navigate these complexities.
Nepal can also take inspiration from initiatives like the Singapore–Rwanda AI Playbook, forging strategic, cross-country partnerships that pool resources and expertise-particularly valuable for smaller states aiming to build a robust AI ecosystem without shouldering prohibitive costs alone.
AI development and adoption require significant resources and dedicated organisations.
Nepal's policy draft mentions public-private partnerships but lacks details on private sector incentives. Countries like India and Singapore drive AI growth through venture capital, incubators and government-backed research. Rather than adding compliance burdens, Nepal should foster AI entrepreneurship with tax incentives, innovation hubs and direct funding. AI-specific startup grants or a venture capital matching scheme could further attract global investment.
AI has huge potential to improve many critical areas such as healthcare, education, employment and finance. However, it comes with equal risk, if not regulated properly. The governance section of the policy is vague in addressing AI risks such as misinformation, surveillance and data privacy. While AI can be used to enhance security and governance, the absence of clear guidelines on ethical AI use leaves room for misuse.
Nepal should ensure legal protections are in place to prevent AI from being exploited for disinformation, financial crimes or intrusive surveillance. While the policy touches on these risks, it does not specify how accountability mechanisms will be enforced or who will be responsible for oversight. Without strong cybersecurity laws, encryption standards and data localization policies, AI-driven platforms can expose sensitive personal and government data to external threats.
To make the AI policy truly effective, Nepal must first establish a robust foundation before enacting sweeping regulations. Instead of viewing AI solely as a governance issue, Nepal should embrace it as a development catalyst-requiring targeted investment, research, infrastructure and workforce training. By prioritising capacity-building over premature regulation, the country can fully harness AI's benefits while responsibly managing its inherent risks. Nepal's AI policy should prioritise three pillars: (1) University-led AI research funding, (2) AI workforce upskilling and (3) AI-driven public-private partnerships.
The authors are associated with Sankhya AI-Kathmandu
