The government of Nepal released the long-awaited National AI Policy on February 4, 2025, for public consultation, following the concept paper on AI in July 2024. However, these offer very little insight into how Nepal as a country intends to approach the evolution of AI in promoting human rights and democratic values based on accountable and trustworthy AI systems, lending itself once again to believe that the country’s policies relating to information and digital technologies continue to skew towards policy-based evidence rather than evidence-based policy.

Technology is undoubtedly continuing to shape democracy. However, democracy has also shaped the legal rules concerning technology. Regardless of the scale of Nepal’s economy, the question of how the country wants to move forward with technology demands an answer. AI technology promises to improve economies, governance and social fabric. It is also predicted to boost labour productivity growth globally by around 1.5 percent, representing almost $7 trillion in extra global GDP over 10 years.

As the AI revolution takes place, policymakers in Nepal should consider what steps to take if the country doesn’t have cutting-edge AI models like those from Anthropic and Open AI, particularly as AI gets increasingly divided in the global order, similar to the digital divide between the haves (countries having both skills and resources to create frontier-level AI models to accelerate national priorities) and the have-nots (all the others).

It would be highly optimistic for Nepal to have its own national AI champions from the onset, requiring policymakers to choose between small, bespoke, lightweight, open-source designs for specific use cases or large enterprise-wide computationally intensive models that may not necessarily be aligned with Nepal’s context. They must also consider whether to use a European or US model, brick-and-mortar data centres or providers in the cloud. Sovereignty and identity will influence these decisions, especially regarding trust in the security of AI running on a foreign-sourced model in a remote data centre.

Human-centred approach

Focusing solely on AI’s economic benefits, greater efficiency, productivity and innovation seems harmful. This strategy risks overlooking key issues such as human rights, social equality and the possible detrimental effects of AI on citizens. Instead, a human-centred approach to AI can focus on human well-being, rights and dignity and ensure that AI technologies are created and implemented to benefit humanity, safeguard fundamental rights and address social issues. It would also prioritise the ethical imperatives required to build confidence in AI systems and protect against adverse outcomes such as bias, discrimination, or privacy infringement.

Leveraging universal guidelines

One of the four pillars of the National AI Strategy is “ensuring responsible and ethical AI development”. In this respect, Universal Guidelines for AI should be considered as a basis for developing an ethical AI policy. At their core, these guidelines protect people by establishing institutional obligations and reinforcing their rights in the context of such systems. It contains crucial guidelines, namely—right to transparency, right to human determination, identification obligation, fairness obligation, assessment and accountability, accuracy, reliability and validity, data quality, public safety, cybersecurity, prohibition on secret profiling, prohibition on unitary scoring and termination obligation.

Since Nepal has endorsed the UNESCO Recommendations on the Ethics of AI, it is recommended that the values and principles included by the organisation should also be used as guidance in developing a national AI policy. The UNESCO Recommendation on the Ethics of AI, adopted in November 2021 by the 193 member States, affirms that “AI actors should make all reasonable efforts to minimise, avoid strengthening or perpetuating applications and discriminatory or biased outcomes throughout the lifecycle of the AI ​​system.”

Guidelines for National Policy

A national AI policy should also incorporate appropriate provisions on generative AI. Given concerns about various human rights implications related to their use, clear guidelines for developing and deploying generative AI are essential.

Strengthening AI Governance for a robust national framework requires establishing an AI governance regulatory body, a national AI policy framework and a national AI risk management framework without which policy becomes orphanised. Foundations for National AI Policy: Proposals for Ethical Governance and Assessment requires the government to establish a High-Level AI Ethics Group/National AI Ethics Commission, develop national AI ethics principles and develop a comprehensive AI assessment framework. These are important building blocks to any successful National AI policy development, and Nepal should not miss this trajectory.

The Ministry of Communication and Information Technology’s commitment to an inclusive policy-making process and encouragement of continuous dialogue among stakeholders to refine and enhance the national AI policy are positive steps. Adopting the recommendations will not only facilitate AI development but also contribute to building public trust and confidence in AI systems, ultimately leading to a more equitable and just Nepali society.

When ingenuity counts for more than brute force, a better way to ensure Nepal’s AI growth would be to attract and retain top researchers from elsewhere. The AI era is still in its infancy, and much remains uncertain. However, the breakthrough AI requires in Nepal will come from giving ideas and talent the opportunity and space to flourish at home rather than copying approaches other countries have adopted.