Harnessing AI, data and technology for growth in Pakistan

Across the world, countries are embedding artificial intelligence in their economic strategies and growth frameworks. From smart health diagnostics to smart traffic systems and taxation, AI can create many opportunities – so how can Pakistan use AI effectively to address its development challenges?

Despite Pakistan’s many growth challenges (including structural deficits in public service delivery, fiscal pressures, and environmental stress), it has a growing digital ecosystem, a large pool of young tech users, and several promising policy pilots. This blog outlines what AI readiness means for Pakistan and identifies areas where AI holds the most potential for transformative change.

Progress and gaps in Pakistan’s digital ecosystem

The latest Government AI Readiness Index by Oxford Insights ranks Pakistan eighth of 17 countries across South and Central Asia – below India, Bangladesh, and Sri Lanka. This index scores countries across several key parameters, including government vision, digital capacity, and data infrastructure. As of 2024, Pakistan was ranked at 97 out of 133 on overall digital infrastructure, skills and usage, and 149 out of 197 on openness of government data, reflecting limited access to official datasets for public use.

Pakistan's core digital ecosystem remains underdeveloped. Administrative data across government agencies such as the National Database and Registration Authority (NADRA) and the Federal Board of Revenue (FBR) remains non-digitised, fragmented, and in silos, and is therefore difficult to integrate and access. Provisions for cross-agency data-sharing protocols are limited. As of 2025, 116 million Pakistanis had access to the internet – an internet penetration rate of 45.7% – implying that over half the population remains offline. 

Despite these gaps, progress has been made in the form of the recently launched National AI Policy 2025, digital skills programmes like DigiSkills, and a growing network of National Incubation Centers (NICs) supported by Ignite, the innovation arm of the Ministry of Information Technology. Pakistan also has a growing number of AI-based startups in fintech, edtech, and healthtech, and is witnessing a growth in IT exports – in 2024-25, these grew by 23.7%.

While challenges remain, there is scope for strengthening institutional capacity, governance structures, and data ecosystems to enable effective AI adoption in Pakistan.

Other countries lead the way in using AI for development 

AI adoption, especially in developing countries, should align with development priorities rather than being driven by technological advancements alone. Examples from around the world demonstrate how technology is being used to address the most immediate policy and development challenges.

• Controlling pollution: In Bangladesh and India, researchers have used machine learning on satellite images to map polluting brick kilns, helping policymakers target inspections and promote cleaner technologies. In Punjab, India, satellite-based fire alerts are being embedded into routine processes to control stubble burning, which remains a major contributor to air pollution in the region.
• Preparing for disasters: Bangladesh and India are also working with Google’s Flood Hub, which provides AI-driven flood forecasts of up to seven days across 100 countries, covering 700 million people. Forecasts in Bangladesh have been linked to anticipatory cash transfers. In the Philippines, the Dynaslope project uses sensors and analytics to provide community-level landslide early warnings.
• Collecting taxes: India’s Goods and Services Tax (GST) system integrates AI for risk detection, identifying suspicious transactions and high-risk taxpayers. South Africa’s Revenue Service (SARS) employs machine learning risk engines, with official reports attributing a third of compliance revenue in some periods to these systems.
• Healthcare interventions: Nigeria and the Philippines are using AI-powered chest X-rays (CAD) to triage tuberculosis cases, with the endorsement of the World Health Organization. São Tomé and Príncipe piloted an AI-guided malaria control programme (Zzapp), which reduced mosquito breeding sites through digitally planned larviciding.
• Improving learning outcomes: India's adaptive learning programme, Mindspark, used technology-aided instruction to boost student scores in math and language in under five months.

What's working in Pakistan? Evidence from IGC-supported pilots 

Research funded by the International Growth Centre  (IGC) demonstrates how tech and data can enhance governance and service delivery. The examples discussed below illustrate the potential of AI across areas such as taxation, environmental governance, urban planning, social protection, and digital inclusion, and present scalable solutions tailored to Pakistan’s needs:

1. Smarter taxation

In collaboration with LUMS, IGC has supported the development of machine learning tools that use satellite imagery to detect unassessed properties and informal sprawl. Piloted in Khyber Pakhtunkhwa province, this work informs cross-verification mechanisms that reduce reliance on manual surveys and improve property tax compliance. A more recent ongoing study in Punjab is testing whether computer vision can improve property tax collection by reducing corruption and increasing fairness. A machine learning algorithm will be developed to predict property valuations from images and then evaluated through a randomised controlled trial (RCT).

2. Air quality monitoring

IGC is working with the Punjab Environment Protection & Climate Change Department (EPCDD) to develop smog alert systems using real-time air quality data and satellite imagery. These tools can be scaled into AI-powered early warning systems for floods, smog, and heatwaves. 

IGC researchers are collaborating with the EPCCD to improve the enforcement of anti-crop-burning policies during the peak smog season. Tools like smoke plume modelling and satellite-based fire detection will power a prototype AI-enabled dashboard, supporting enforcement officers and Union Council-level coordinators in real-time monitoring. The system will also be linked with drone surveillance data and field validation, offering a scalable model for subnational environmental governance.

3. Urban planning

Ongoing IGC-supported research in Lahore is using Call Data Records (CDRs) from telecom networks to map migration flows into cities driven by socio-economic pressures and climate shocks. By applying AI and GIS tools, researchers can identify commercial clusters, infrastructure gaps, and under-taxed high-activity zones. This real-time mapping of the informal urban economy can directly inform zoning reform, congestion management, and more inclusive urban policy.

4. Targeting vulnerability 

During the COVID-19 pandemic, IGC research in Karachi leveraged a geo-spatial targeting tool for social protection using electricity consumption data from 2.4 million households to map vulnerability with 70% accuracy. Combined with neighbourhood mapping and satellite-derived density, the tool linked poverty with disease spread risk. Similar efforts in Punjab are integrating survey data, remote sensing, and CDRs to create high-resolution poverty maps for better social protection targeting.

5. Digital inclusion

IGC is piloting a programme with the Kaarvan Crafts Foundation, Daraz, and the Punjab Social Welfare Department to assess how generative AI tools can boost productivity for microenterprises, especially women-led businesses. The study compares AI training with Google-based training across 600 participants, focusing on applications such as customer service and bookkeeping. This research builds on IGC’s work on digital inclusion and aims to generate scalable evidence on the role of AI in supporting informal businesses.

How can Pakistan move towards the systemic adoption of AI?

1. AI strategy should be anchored in development goals

In July 2025, Pakistan’s cabinet approved the National AI Policy, signalling a strategic commitment to mainstreaming AI in governance and development. The policy sets ambitious goals – including training AI professionals, launching civic AI projects and homegrown AI products, establishing an AI Council, and creating a dedicated innovation and venture fund. It also emphasises cybersecurity, ethical use, and data sovereignty. While the policy provides a strong framework, AI must be directly tied to development outcomes like tax compliance, smog mitigation, and inclusive service delivery, rather than abstract targets.

2. The government should become a lead user of AI

Pakistan must shift from being a passive adopter to a lead user, embedding AI into everyday workflows. This means integrating AI not as a one-off innovation, but as a core part of everyday public sector workflows from forecasting school dropouts to optimising public health outreach.

3. Invest in building human capital beyond coding

It isn’t enough for Pakistan to have coders; it needs professionals who combine domain expertise with technical skills. Programmes like the National Vocational & Technical Training Commission and DigiSkills must focus on sectoral applications to health, agriculture, or education. At the same time, the government must invest in data protection, transparency, and ethics. 

4. Ensure a robust data ecosystem and responsible access

Most Pakistani agencies hold siloed and inaccessible administrative data, limiting the use of AI in policymaking. IGC’s work has shown the value of linking remote sensing with tax, land, and social protection records to improve targeting. Going forward, Pakistan must develop anonymised, interoperable data platforms with strong safeguards to enable secure access for government and researchers.

5. Strengthen the private sector and research environment

AI’s biggest innovation potential lies outside government. Startups in fintech, healthtech, and edtech need secure data, sandbox regulations, and public-private partnerships – institutions like Ignite and NICs can catalyse this if aligned with adaptive regulation. 

6. Build sector-specific capacity for AI adoption

Finally, Pakistan’s digital transition requires professionals trained not just in data science, but also in applying AI to public service domains such as health, law, climate, and urban planning. Embedding these skills into higher education curricula and professional training programmes will help ensure AI is used to solve real-world development problems.

How can the IGC help government and policymakers in Pakistan? 

1. Help generate evidence and use cases

IGC uses rigorous methods (RCTs, quasi-experiments, diagnostics) to evaluate the effectiveness and limitations of AI tools in the public sector. This evidence helps governments and donors decide what to scale, adapt, or redesign, reducing risks of wasted investment or unintended outcomes in tech-driven reform.

2. Build institutional capacity

IGC works directly with line departments (such as the Environment Protection Department or the Punjab Social Protection Authority) to embed AI pilots into government workflows, ensuring tools are not externally imposed but internally understood and used. This includes working with mid-level bureaucrats, improving data access protocols, and supporting the development of in-house prototypes and analytical capacity, laying the groundwork for broader institutional uptake.

3. Link research and policy ecosystems

IGC supports stronger ties between startups, researchers, and government institutions, enabling co-creation of solutions and smoother adoption. AI pilots on tax, smog, and microenterprise development are often co-developed with public agencies and local partners. By bridging innovation ecosystems and public policy, IGC helps shape more adaptive and responsive governance systems.

Explore IGC Pakistan