Submit your Article

In recent years, traditional media houses across the globe have resorted to layoffs as a last-ditch effort to stay afloat; if not for the long term, then at least for a few more years. This wave of downsizing began during the Covid-19 crisis and has yet to subside.

Media organizations are now restructuring into smaller, smarter and more agile newsrooms to cut costs. They are grappling with a severe financial crisis as conventional revenue streams dry up and new ones are slow to emerge. Nepal is no exception to this trend. To reduce expenses, many media houses are working to merge operations across print, radio, television and digital platforms into unified newsrooms. The only seemingly viable, though not well-thought-out option has been to scale down operations to match dwindling revenues. 

In this context, a wide range of Artificial Intelligence (AI) tools could prove to be a boon for the fragile media landscape, potentially helping to fill gaps left by staff reductions. However, before embracing AI more broadly, it is crucial for media houses to formulate clear policies to ensure its ethical, transparent and effective use.

While some media houses have already started using AI tools, their applications remain minimal and largely unregulated. It is high time media houses moved decisively, from the Gutenberg-era newsroom to an AI-equipped, high-tech newsroom. A key first step in this transition is to provide training for journalists and collaborate with technology companies to develop customized newsroom tools. While the adoption of AI is not without costs, it can be a cost-effective alternative in the long run, gradually replacing outdated editorial structures.

At present, AI use in Nepal’s newsrooms is limited to individual journalists. Many AI-generated, translated or edited texts are published without any editorial supervision. While no comprehensive study has been conducted to assess the use of AI in Nepali newsrooms so far, a recent survey by Rajiv Timalsina, a student of Kantipur City College, provides some insights. According to the survey, 38 percent of journalists use AI tools for transcription tasks such as documenting interviews. Around 22 percent use them for fact-checking through platforms like Google Fact Check or image verification engines, while only 18 percent use audience analytics tools to understand reader behavior.

The next step after introducing AI tools is to train employees to use them effectively, particularly to improve the quality of writing and editing. Currently, there is a lack of trained human resources in the newsroom, and local journalists are still in the early stages of AI adoption.  While some non-governmental organizations have begun offering training, there has been little to no institutional collaborations.

Media houses must establish dedicated AI departments and AI editors to provide proper insight and guidance. Without this, the unchecked use of AI could lead to serious problems. If possible, Nepali media should also seek collaboration with international media organizations to learn from their experiences, though even global media outlets are still experimenting with AI integration.

In 2024, The New York Times publicly released a document outlining its approach to AI in the newsroom. The US media company said it does not use AI to write news or articles. It said it uses AI in three main ways: as a tool in the service of its journalistic mission, under human guidance and review and transparent use. Compared to other international media outlets, it has adopted a more cautious stance on AI use, maintaining that human creativity remains central to content creation.

In 2023, The Financial Times appointed Madhumita Murgia as its first AI editor. The following year, The New York Times rolled out its first generative AI features for subscribers. The same year, The Washington Post launched “Ask the Post AI”, which it described as a generative AI tool leveraging the publication’s deeply-sourced, fact-based journalism to deliver summary answers and curated results directly to users. 

In neighboring India, The Hindustan Times joined the AI race in 2024, establishing a 15-member team to work on GenAI-based initiatives such as news bots, personalization, audience engagement, monetization and subscription strategies. 

In Nepal, while journalists have begun using publicly-available AI tools, institutional adoption remains limited. However, some media outlets have started integrating AI technologies in various ways, from digital news readers to audio transcription, translation, image and text generation, and even news writing.

Providing summaries alongside news articles, with editorial endorsement, is a growing global trend. Onlinekhabar is among the Nepali media following this trend. Annapurna Post has also taken help of AI for its digital reader tool.  However, some news outlets, which lack strong editorial oversight, are publishing AI-generated summaries that are flawed or misleading.

With the use of AI at the individual level growing, media houses must ensure that AI is used responsibly and ethically. This is necessary both to maintain editorial integrity and to earn people’s trust. With strategic investment and collaboration with tech companies, AI could unlock new opportunities for Nepali media houses. 

AI tools can help summarize news stories, but editors must have the final say. The New York Times’ own experience shows that AI-generated summaries often fail to fully capture the nuances of original articles. Many believe that human-written summaries are still superior. Despite this, the US media company’s AI team has been refining its tools, acknowledging that while AI is not perfect, it can still help free up editorial staff for other important tasks. 

With the right human guidance, generative AI can also be used to create visuals to accompany new stories. Some Nepali media houses have started using AI-generated images, but these are only accurate when journalists provide detailed guidance. Otherwise, there is a risk of misrepresentation and factual inaccuracies. 

Resource constraints have long prevented Nepal media from producing investigative, analytical and in-depth news stories. AI could help bridge this gap. It can assist in scanning documents, analyzing data and identifying leads for investigative reports. These are the tasks that journalists often struggle to manage under tight deadlines. AI can also support wider and more efficient coverage by translating news into multiple languages to reach broader audiences. Some outlets in Nepal have already begun experimenting with this approach.

AI tools can be used to support news writing and editing. However, this should always be done under direct editorial supervision. These tools are best used to prepare preliminary drafts. For journalists, AI can help identify trending topics, suggest potential sources, summarize lengthy documents, conduct background checks and even engage audiences more effectively.

While investing in AI infrastructure may place an additional financial burden on media houses in the short term, it could prove vital to their long-term survival. On one hand, AI can significantly enhance the quality and efficiency of news production; on the other, a compact AI-powered newsroom can help reduce human resource costs. Compared to other countries, Nepali media remain behind in adopting technology. But the use of AI in newsrooms is no longer a distant possibility, it is a present-day reality.  The question is not whether to use AI, but how to use it effectively to harness its benefits.

The fiscal year 2024–25 marked a cautiously optimistic phase in Nepal’s post-pandemic economic recovery. With a projected GDP growth of 4.61 percent, a narrowed fiscal deficit and record foreign exchange reserves, Nepal demonstrated notable resilience. However, beneath these surface indicators lies a complex interplay of structural weaknesses, external dependencies and opportunities that deserve closer scrutiny.

What it really means

At first glance, Nepal’s GDP growth of 4.61 percent appears moderate and consistent with a recovering economy. But this figure, while respectable, remains below the 7–8 percent growth rate necessary for rapid poverty reduction and meaningful job creation. The marginal increase in growth from the previous year’s 3.9 percent suggests a slow recovery rather than robust expansion.

More importantly, much of this growth was consumption-led and driven by remittance inflows, rather than investment-led industrial or export expansion. This signals a structural concern: Nepal’s economy continues to lean heavily on external income rather than internal productivity.

Services dominate, industries lag

The composition of GDP reflects deep-rooted imbalances. The services sector contributed over 62 percent to GDP, dwarfing agriculture (25.2 percent) and industry (12.8 percent). While services growth—particularly in transport, storage, and financial activities—is encouraging, it raises questions about sustainability. Services, especially low-productivity informal ones, often expand when there is a lack of industrial dynamism.

The industrial sector, despite moderate growth in construction and manufacturing, remains constrained by infrastructural bottlenecks, power reliability issues and limited domestic and foreign investment. Agriculture, although vital for employment, continues to suffer from low productivity, climate vulnerability and lack of commercialization.

A silver lining?

Headline inflation was 4.72 percent, down from previous years. This reflects effective monetary tightening and better supply chain management. However, food inflation persisted around 3.3 percent, affecting poor households disproportionately.

More analytically, the disinflationary trend owes much to suppressed demand and import-based consumption rather than domestic supply resilience. In a context where inflation in neighboring India remains high, Nepal’s price stability is fragile due to the currency peg and trade dependence. Any external price shock—especially in fuel or food—could reverse the gains swiftly.

Strength built on vulnerability

Remittances grew by 9.4 percent, reaching over Rs 1trn. On the surface, this is a strong signal of income support for households and foreign exchange stability. However, the economy’s growing reliance on labor exports (over 25 percent of GDP) reflects domestic weaknesses in job creation. Migration is not a sign of strength—it is often a symptom of failure to absorb labor at home.

The surge in exports (up 57 percent) is driven by a few commodities like edible oil re-exports and textiles, making it highly sensitive to global demand and bilateral trade policies. The trade deficit remains wide, and Nepal continues to import high-value goods while exporting low-value products—an unsustainable model.

The record-high foreign exchange reserves (covering over 14 months of imports) are welcome but largely attributable to remittances and restrained import demand rather than export competitiveness.

Improved discipline, but at what cost?

Nepal’s fiscal deficit declined sharply—from Rs 70bn to around Rs 16bn in the first eight months—thanks to higher revenue growth and restrained spending. While this reflects improved fiscal discipline, a closer look reveals underperformance in capital expenditure. Many development projects remained delayed or underfunded due to bureaucratic inefficiency, procurement issues and political instability.

Moreover, public debt is at 43.8 percent of GDP—moderate by international standards—but its composition is shifting toward more domestic borrowing, raising concerns over future interest liabilities and crowding out of private investment.

Loosening sans uptake

The Nepal Rastra Bank lowered policy rates to inject liquidity into the economy, leading to historic lows in lending rates. Yet credit uptake remained sluggish, indicating low investor confidence and weak private sector appetite for expansion. The rise in non-performing loans to 4.9 percent underscores emerging stress in the banking system, which could worsen if economic recovery remains tepid.

This disconnect between monetary easing and private sector response suggests deeper structural barriers—legal hurdles, creditworthiness concerns and weak project pipelines.

Climate shocks and structural risks

Nepal’s economic resilience was tested by major floods in mid-2024, causing damage equivalent to 0.8 percent of GDP. This highlights the increasing economic cost of climate change, especially for a country with fragile topography and inadequate disaster preparedness. Yet, climate adaptation and green investment remain minimal in budget allocations.

Additionally, long-term risks—including heavy remittance dependence, trade imbalances, political instability and underemployment—remain unaddressed. These challenges, if not structurally tackled, could stall Nepal’s path to middle-income status.

Conclusion: Resilient, yet restricted

Nepal's economic performance in 2024–25 reflected stability without transformation. The country avoided crisis and managed moderate growth, but it did not make the leap toward a more productive, inclusive or diversified economy. The gains were largely reactive rather than strategic—buoyed by remittances, import compression and fiscal restraint rather than innovation or competitiveness.

To transition from recovery to take-off, Nepal must move beyond short-term fixes. Reforms in public administration, industrial policy, export diversification, education and climate resilience are essential. Without them, the economy risks settling into a low-growth equilibrium marked by dependence, inequality and untapped potential.

With the rise of generative AI in research and education, a question keeps coming to mind: How is the way we understand knowledge changing as AI becomes a bigger part of our daily learning and work? This is not just a question for academics or tech experts; it affects everyone who relies on knowledge to make decisions, express ideas or contribute to their communities. We are at a point where the very act of knowing is changing—not just how we know, but who we consider to be the "knower." When a machine writes an article, summarizes a book, or helps design a curriculum, what role does the human thinker still play?

On the one hand, this technology opens up new possibilities. A student in a remote village in Nepal can now access summaries of global literature, translate complex theories into Nepali or get help writing a research paper—all at the click of a button. Generative AI can be a powerful tool for breaking down barriers of language, access and time. On the other hand, there’s also the risk that we may stop thinking for ourselves, relying too heavily on a tool that reflects patterns, not true understanding. In a world where so much is automated, what happens to reflection, to critical thought, and to the slow and sometimes uncomfortable process of finding our own insights?

As I struggled with these questions, I found some guidance in Eastern philosophy. While ancient texts didn’t predict AI or digital tools, they did take the question of knowledge very seriously. In the Eastern tradition, knowledge (jñāna) is not just about gathering facts. It’s something that transforms us, something that reveals the self, the world, and the relationship between the two. Importantly, it is always tied to ethics. One does not seek knowledge simply to win arguments or impress others; knowledge is pursued to live rightly, act responsibly and move closer to truth and liberation.

This is especially relevant now as generative AI begins to influence how we write, research and think. The Upanishads tell us that the student should not just ask, “What is this?” but also, “Who am I?” It’s a question of identity, intention and inner clarity. When I use AI to write a paragraph or generate ideas, I try to stay aware of what part of me is involved. Am I using the tool to clarify my thoughts or to avoid doing the hard work of thinking? Am I driven by curiosity or by convenience? These may seem like philosophical questions, but they have very practical implications. Imagine a college student in Kathmandu working on their assignments. With AI, they can generate drafts in minutes, find sources and even correct their grammar. But if they stop reading, stop questioning and simply copy what the machine offers, they may submit a polished assignment—but miss the point of education entirely.

The machine can assist, but it cannot reflect. It cannot care. It cannot ask, “Is this meaningful to my society, my values, or my life? Eastern philosophy offers a helpful metaphor here: the yantra or instrument. Tools are nothing new. Humans have always used tools to extend our abilities—whether it’s the plough in agriculture, the loom in weaving or the telescope in astronomy. What matters is not just the tool, but how we use it, and for what purpose.

The Bhagavad Gita reminds us that the right action must be performed without attachment to the outcome, guided by clarity and duty—dharma. In today’s world, AI is a new yantra, but it requires the same discipline. We must ask: is it helping me fulfill my role as a student, researcher or a citizen? Or is it just making things easier at the cost of meaning? This doesn’t mean we should fear technology. Far from it. Used wisely, generative AI can become a partner in learning, a bridge across educational gaps and a tool to preserve and even regenerate local knowledge.

Imagine AI trained to document indigenous languages in Nepal or to translate oral histories into written texts. Imagine teachers using AI to create personalized learning experiences for students from different backgrounds and needs. These are exciting possibilities—but they can only become a reality if we use them with care, ethics, and awareness.

 In Eastern philosophy, ethics is not separate from knowledge. Truth (satya) is not just about factual correctness; it is about aligning what we know, say and do. When we conduct research with the help of AI, it still matters that we acknowledge our sources, credit others and question the biases embedded in the tools we use.

It still matters that we ask: Does this help society? Does it deepen understanding? Or am I simply using a machine to do my work for me? This brings us back to the idea of rethinking how we understand and interpret knowledge. Perhaps the real shift is not just technological—from books to machines, from human writers to AI—but ethical.

It is a turn toward remembering that knowledge is not neutral. It shapes lives, it holds power and it demands responsibility. In this light, AI is neither a savior nor a threat. It becomes a mirror, reflecting our habits, assumptions and goals. And it asks us: What kind of knowers do we want to be?

In a country like Nepal, where tradition and modernity often walk side by side, we have a unique opportunity. We can engage with new technologies not blindly, but with the wisdom of our philosophical traditions.

We can teach students not just how to use AI, but how to think with it—critically, ethically and reflectively. We can build an academic culture that values not just output, but insight. In the end, Eastern philosophy doesn’t reject tools. It simply reminds us: We must be worthy users of them.

The Annapurna Express’s recent coverage (June 11 & 15, 2025) outlines Nepal Rastra Bank’s (NRB) paradigm shift toward inclusive policy making alongside pointed critiques of financial sector oligopolies under Governor Biswo Poudel’s stewardship. While these dual initiatives ostensibly represent a break from traditional monetary governance paradigms, their transformative potential remains circumscribed by institutional implementation constraints. The Governor’s grassroots consultations with entrepreneurs signal a welcome democratization of policy formulation, particularly regarding climate-adaptive financing and (small and medium-sized enterprise) SME sector needs. Parallel criticisms of credit concentration among privileged business houses and individuals during parliamentary debates on financial sector reform legislation reveal acute awareness of systemic inequities. However, mere articulation of these concerns proves insufficient. There is a need for technically sophisticated policy instruments, rigorous monitoring mechanisms and enforceable regulatory safeguards. The fundamental challenge confronting NRB transcends rhetorical commitments, residing instead in its institutional capacity to convert participatory inputs and diagnostic critiques into measurable policy outcomes. Without such operational competence, these ostensibly progressive measures risk remaining performative gestures rather than effecting substantive financial sector transformation. The central bank’s ability to institutionalize technical implementation frameworks will ultimately determine whether this reorientation represents genuine reform or merely cosmetic governance adjustments.

Advisory mechanism and institutional redundancy

Governor Poudel’s recruitment of a three-member advisory committee underscores structural inefficiencies within NRB governance. Despite NRB’s existing cadre of internationally trained professionals, reliance on this external back door entrants raises concerns about internal confidence and the potential for political patronage. Advisory bodies often serve symbolic rather than substantive roles, diluting accountability and diverting attention from rigorous and data-driven analysis. The sidelining of internal technical expertise in favor of ceremonial consultations undermines efforts to implement advanced financial modeling and scientific risk assessments. To move forward, NRB must strengthen in-house analytical capacity and deploy modern surveillance technologies, transitioning from a bureaucratic institution to a knowledge-driven central bank.

 Implementation vs. preparation

Governor Poudel’s technocratic background and field engagement are commendable, yet policy effectiveness hinges on implementation. With inflation at 6.05 percent in mid-December 2024 and productive sector lending stagnant at 15.2 percent of total credit, Nepal faces the challenge of balancing price stability with growth imperatives. The NRB's inflation target of 6.5 percent needs to be harmonized with the pressing requirement for funding climate-smart infrastructure and support for SMEs. However, the central bank’s autonomy remains fragile, persistently challenged by fiscal dominance and political interference. Cross-country evidence underscores that monetary policy effectiveness depends on institutional independence, a standard NRB struggles to consistently meet.

 Monetary policy’s developmental role

Monetary policy in Nepal serves a critical dual function: stabilizing the macroeconomy and enabling structural transformation through resource allocation. Yet, its impact is limited by an underdeveloped financial sector and the predominance of informal credit markets and remittance inflows, which distort price signals. Contractual savings institutions such as the Employment Fund, Social Security Fund, and Citizen Investment Trust, including police and army funds, manage significant public savings. But they operate with minimal regulatory oversight. This gap fosters risky lending practices and threatens financial stability. NRB must integrate these entities into a comprehensive regulatory framework, aligning their operations with macroeconomic objectives. Furthermore, NRB should enhance oversight of primary and secondary capital markets, where rooted rent seeking interest groups undermine market integrity. Transparent regulation can bolster SME financing and investor confidence. Public-private credit risk-sharing models, SME financing and fintech platforms offer promising avenues to expand investment opportunities and financial inclusion.

Risks in financial innovation

While securitization offers liquidity and risk diversification, in weak financial systems it can amplify systemic vulnerabilities. Instruments like collateralized loan obligations (CLOs) and collateralized debt obligations (CDOs) often misallocate risk and encourage moral hazard. In Nepal’s nascent capital markets, such products risk heightening financial instability rather than mitigating it. The NRB should enforce risk-retention mandates, strengthen disclosure protocols, and develop sector-specific risk models to ensure financial innovation reinforces rather than undermines systemic resilience.

 Structural deficiencies in financial architecture

Three critical deficiencies weaken Nepal’s financial system. First, inflation measurement relies on outdated commodity baskets, failing to capture actual household expenditures in key sectors. Updating consumption weightings is vital for credible policymaking. Second, credit allocation failures, especially in state-owned banks, lead to mispriced risk and inefficient capital flow, evident in speculative hydropower lending and trade finance manipulation. Implementing risk-based lending and strengthening governance are urgent. Third, financial exclusion persists due to legacy banking practices that marginalize SMEs, pushing them toward informal credit markets. Hybrid credit models and dedicated SME facilities could help bridge this financing gap.

 Curbing NPLs and strengthening risk oversight

Nepal’s escalating non-performing loans (NPLs) underscore systemic vulnerabilities in credit underwriting and risk management. According to the NRB’s Financial Stability Report (2024), the aggregate NPL-to-total-loan ratio surged to 3.86 percent, marking a 3.4 percent year-on-year increase, with total distressed assets reaching Rs 199.66bn. Alarmingly, over 56 percent of these NPLs are classified as “loss” assets, reflecting severe deterioration in asset quality. Sectoral analysis reveals acute stress in construction (7.28 percent), followed by fisheries (6.65 percent), agriculture (6.22 percent), and metal production (6.09 percent). The banking sector’s NPL composition has worsened, with loss-category loans now dominating at 56.66 percent. Despite their limited GDP contribution, loan portfolios remain disproportionately concentrated in consumption and retail sectors, exposing financial institutions to unproductive risk. The absence of granular, sector-specific risk assessments particularly in high-exposure sectors like hydropower heightens systemic fragility. To preempt further instability, the NRB must transition from reactive oversight to predictive risk analytics, implementing rigorous stress-testing frameworks and Likert-scale credit scoring models tailored to individual BFIs and each sector. Current reliance on narrative-based advisory mechanisms, devoid of empirical validation, perpetuates cyclical vulnerabilities. Without institutionalizing data-driven risk analysis and surveillance, Nepal’s financial sector risks replicating crises, necessitating urgent reforms in supervisory methodologies to align credit allocation with sustainable economic priorities.

 Takeaway

The NRB’s recent policy shift toward inclusivity and evidence-based decision-making reflects a theoretically progressive stance, recognizing the need for broader economic and investment equity. However, rooted structural weaknesses such as institutional inefficiencies, outdated metrics, and governance gaps undermine substantive reform, risking purely symbolic change. The disproportionate focus on price stability over equitable growth highlights a misalignment with Nepal’s developmental needs, necessitating integrated strategies prioritizing investment in micro, small, and medium enterprises, rural finance, financial inclusion, and sectoral productivity. Effective implementation hinges on preserving technocratic autonomy, as NRB independence is foundational to credible governance. The upcoming monetary policy will test the NRB’s capacity to convert rhetoric into data-driven action, balancing investment expansion with regulatory rigor against politically connected elite businesses capture. As the NRB transitions from macroeconomic stabilizer to proactive architect of Nepal’s economic future, Governor Poudel’s leadership must be judged by tangible outcomes, policy coherence, execution efficacy, and measurable progress across socioeconomic strata. This demands both analytical sophistication and institutional resilience.