Legal Blueprint for Artificial Intelligence and the Nepali Techonomy

The release of OpenAI in 2015 and Chat GPT in 2019 represents significant moments in the growth of Artificial Intelligence (AI). However, this advancement is less sudden than we think. AI has ancient roots that date back to as far as the Greek myths. For example, the story of the Greek god Hephaestus and his creation, Talos, a bronze automaton created to protect the island of Crete exemplifies this fascination with the creation of automation. Further east, we can even find similar links within Vedic science as well. In the creations of Vishwakarma, the god of architecture, we often find imbued maya, the idea of consciousness in machines. Additionally, iIn the mortal context, this entracement dates back to the Renaissance period when Da Vinci explored the nexus of homo vitruvius (perfect man) and homo artificialis (artificial man). 

However, while this fascination has continued for time immemorial, the foundation of AI was only laid in 1950 when Alan Turing started exploring the fundamentals of machine intelligence. This led to the development of AI as we know it, as shown in Table 1. Today, AI has manifested into an ever-growing pervasive technology. AI’s rapid growth can be seen through the lens of Moore’s Law, which states that computers become faster, cheaper, and smaller over time. However, this exponentially improving technology has also led to many questions regarding its ethical implications and its contribution to the widening north-south digital divide.

Key events in AI history

The Imperativeness of AI Laws in Nepal

In order to mitigate the growing concerns mentioned above, it is imperative that governments must prioritize the creation of relevant laws. As evidenced by the AI readiness index, Nepal is far behind in this effort as it ranks 150 out of the 193 countries, similar to other developing countries. This is because of increasing challenges to develop and maintain a robust digital public infrastructure system. The index, which assesses the AI readiness of governments across the world shows that there is a significant digital divide between the global North and the global South (shown in Figure 1 and 2) which reflects present economic inequalities. Thus, as part of a proactive approach to governance, laws to govern AI must help regulate the AI market while promoting its ethical use and fostering innovation. For this, one important method is to take inspiration from preexisting laws.

Lessons from Evolution of Laws

The Nepali legal system has been siphoning wisdom from a range of different sources which has come to become a hybrid legal system over the past 50 years. Giving continuity to this dynamic practice can be resourceful for Nepal due to the positive correlation between codification of laws and development. 

It is common for present laws to be derived from historical international precedents. For instance, the derivation of human rights is embedded in the French Declaration of the Rights of Man. The same can hold true for AI laws and regulations as well. One example to take inspiration from is the Napoleonic Code’s emphasis on legal codification and clarity, uniformity and consistency, equality, and non-discrimination, as well as secular and objective standards. These principles can guide the development of clear, consistent, and fair AI regulations, especially for transitioning economies like Nepal. Thus, as the government moves forward beyond the AI strategy concept note (released in July 2024) and towards the creation of an AI Act, it must take into account international best practices. This will help introduce certainty and control over the ever-changing AI ecosystem leading to better economic development.

Fig 1: Average AI Readiness by Region | Visualizing the digital divideData: Oxford Insights |AI Readiness Index 2023

Fig 2: Average AI readiness global north v/s global south | Visualizing the digital divide Source: Oxford Insights | AI Readiness Index 2023

Exploring AI Laws in Nepal: Risk-Weighted, Outcome-based Recommendations

In addition to taking inspiration from best practices, Nepal’s IT laws should also be a combination of risk-weighted and outcome-based approaches in order to nourish the AI ecosystem. So far, Nepal’s IT laws are notorious for not being supportive of the private sector due to their blanket approaches which either leaves security risks or stifles innovation. 

Risk-weighted laws involve tailoring regulations based on the potential risk associated with specific AI applications. This means that classifying high-risk areas such as data use in healthcare should have stricter regulations and low-risk areas such as data use in song recommendations should have a more relaxed version of those regulations. On the other hand, outcome-based laws focus on achieving specific goals, such as ensuring AI systems are unbiased, transparent, and fair. The combination of these two approaches helps balance innovation and regulation, increasing the relevance timeline of technology-related laws. Keeping this in mind, the following are some areas that Nepal’s budding AI legal framework should consider. 

• Regulation of Large Language Models (LLMs): Large Language Models like GPT-4 are intelligent systems able to understand and generate human-like text. A popular example of this is ChatGPT, in which Nepal’s interest stands in second place in terms of Google searches. Given this popularity within the Nepali market, it leaves us vulnerable to the myriad of risks LMMs bring with them. The biggest risk they pose is their precarious production of biased knowledge and misinformation. Thus, regulatory efforts focused on systematically and regularly monitoring and updating models with the most recent data can minimize the threat posed by LLMs. 

• Transparency: In the growing creative industry of Nepal, saturation of AI content poses risks for human-produced content. Additionally, emerging deepfakes blur the lines between fiction and reality and can be vastly misused to generate misinformation. In this case, clarification of the means of production such as mandatory labeling of AI-generated content would help ensure a transparent and accountable outcome by distinguishing between the AI and human content realms. 

• Supporting innovation: Laws and regulations should foster the technology industry in Nepal due to its high yield and high benefit possibilities for developing countries. This should be done by incentivizing desired action to promote transparent, fair, and non-discriminatory AI use. Incentives can include tax subsidies as benefits to reducing labor costs of innovation. Additionally, financial contributions for the development of startups can help support emerging tech companies to establish their roots. 

Adaptability: In the midst of the 4th industrial revolution driven by AI, laws and regulations should prioritize human centricity or needs, thereby increasing their adaptability. This can be done by adopting a technology-neutral definition of AI, which does not discriminate between emerging technologies and established ones. It rather allows for market mechanisms to choose which is the most cost-effective alternative. This also provides consistency over AI regulations and promotes fair, accountable and inclusive outcomes which in turn future-proofs these laws.

Outlook 

The concept note on AI released by the Ministry of Communication and Information Technology (MoCIT) is a key step to mitigate the north-south digital divide. It welcomes a structure that includes AI laws and regulations in its meticulously planned timeline which helps us realize that economic equity isn’t a notion that Nepal can only dream of. This opens Nepal up to a spectrum of economic incentives through tech startups and the incorporation of AI in government systems. This technology-supported, human-centric economic growth is potentially the fulcrum for a smooth LDC graduation. However, it requires significant efforts to be seamless. The development of AI laws and of digital public infrastructure are complementary to each other to support innovation and economic growth, explore essential ethical considerations, and effectively utilize AI for Nepal to channel its benefits. The recommendations in this article serve as the blueprint for AI laws to cultivate a conducive legal environment for a tech-empowered economy. In this way, Nepal can catch up with speedy AI developments around the world.