For decades, taxation and social welfare systems have been looped within a bureaucratic time warp – buried under paper trails thicker than your morning newspaper. Tax evasion? Rampant. Welfare fraud? A logistical nightmare. While tax collectors chase down elusive billionaires with offshore accounts and clever loopholes, essential welfare funds often disappear into a black hole of administrative blunders and outdated systems.
The urge to transform our fiscal nightmares into a series of streamlined processes is underscored by the harrowing realities of income inequality. With the richest 10% controlling over 60% of the country’s wealth and the bottom half holding onto just 11.2%, the gap is both glaring and continual. Meanwhile, Pakistan’s tax-to-GDP ratio remains stagnant at a mere 9.2%, the lowest in the entirety of South Asia, while approximately PKR 3.2 million also goes down the drain on account of lost track revenue annually. Thus, Pakistan’s economic cohesiveness is plagued by digital, structural, and alarming institutional problems.
Consider the potential of artificial intelligence: tools that can detect tax evasion with unprecedented speed and accuracy, and systems that help ensure welfare benefits are delivered efficiently to the rightful recipients, minimising leakage and misuse of public funds. With proper implementation, AI could transform Pakistan’s tax system into a fair, fraud-resistant machine. Without it, tax evasion might just remain the country’s favourite national sport, with a potentially greater record of trophies than cricket.
In a meeting in 2024, the Prime Minister ordered the Federal Board of Revenue to uncover individuals accumulating a high net worth and to put a stop to income leakage. And it’s safe to say the FBR paid attention. Diving into the reforms, firstly, Pakistan saw a whopping 30–35% boost in sectoral tax revenues. How? Through the implementation of digital QR codes to effectively monitor the movement of goods in sectors like sugar and tobacco. Secondly, a digital tax filing platform, IRIS 2.0, is currently being enhanced to include automatic audit prompts and taxpayer analysis focusing on behavioural variations. Lastly, the Automated Risk-Based Audit System has also substituted unreliable manual audits with risk scores generated through machine learning. Moreover, AI has been doing its detective work efficiently. In collaboration with NADRA, the FBR utilised integrated data streams and successfully identified 4.9 million unregistered taxpayers and 1.35 million non-filers.
It is not AI that we should fear but cunning policies cloaked as innovation
However, AI does not just revolve around the popular anti-capitalist slogan of ‘eat the rich’ or is limited to acquiring tax payments from the elite. Welfare delivery is being reshaped, too. With algorithmic scoring, the Benazir Income Support Programme (BISP) cut fraud and exclusion errors by 22%. Initiatives like SOS-Pak are also involved in utilising real-time poverty heat maps and allowing policymakers to channel targeted aid. Similarly, AI-driven video analytics in sugar mills resulted in an over 20% decline in production underreporting. The takeaway? Everything is digitally monitored under the eyes of AI.
However, before we get too excited, there is a very significant question. What if the algorithm gets it all wrong?
Ironically, in 2021, the FBR’s own servers were hacked – indicative of Pakistan’s weak data protection and privacy mechanisms. Without proper safeguards, personal information like CNICs and bank account details can be jeopardised. Algorithmic bias poses yet another threat. AI tools primarily trained based on urbanised data have a high likelihood of excluding minorities, women, and informal workers or wrongly flagging them for audits. In a developing nation like Pakistan, transparency, consent, and human oversight are imperative to avoid an ‘AI dystopia’.
So what’s next for Pakistan’s fiscal frontier? Since its AI dreams seem too far-fetched, given its personal tech capacity.
Pakistan has repeatedly relied on global consultancy firms to steer its digital transformation – from providing bias-detection toolkits to algorithmic audits and hybrid decision-making models. Notably, the role of firms like McKinsey or UNDP has extended far beyond traditional advisory services, aiding the FBR and Pakistan Ministry of Finance to bolster AI literacy, establish internal data science units and develop predictive models – all while ensuring that sensitive fiscal and welfare data remains securely in-state.
However, you can’t just toss AI into a system built for dusty filing cabinets and hope for certain miracles. It is non-negotiable to pass a robust Data Protection Act to safeguard against misuse and mandate regular algorithm fairness audits to catch discrepancies before they bite. Not only this, but AI tools must be localised regarding chatbots in various languages, and these should be deeply embedded in a governmental framework, ensuring a savvy mixture of automation and human oversight. Pakistan must commit to a long-term strategy that fuses smart technologies with smarter policies, ensuring that AI serves as a tool for justice and not just efficiency.
Because in this day and age, it is not AI that we should fear but cunning policies cloaked as innovation.