From Reports to Results: AI in Public Sector Governance

Pakistan has spent years trying to improve the performance of its public sector enterprises. Boards have been reconstituted, reporting requirements strengthened and new governance frameworks introduced. Yet financial losses, operational inefficiencies and weak accountability remain persistent concerns. This raises an important question: if problems are already documented in financial statements, audit reports and performance reviews, why are they so difficult to resolve?

Perhaps the problem is not a lack of information, but the failure to turn information into timely decisions and corrective action.

Artificial intelligence (AI) could help bridge this gap, provided it is used to improve governance rather than simply produce more reports.

Most boards still rely on scheduled meetings, management presentations and periodic financial reviews. These remain necessary, but problems can develop long before they reach the boardroom. By then, opportunities for early intervention may have been lost.

AI offers a different approach. Through continuous monitoring and predictive analysis, boards could receive early warnings about unusual expenditure, deteriorating financial performance, procurement risks and operational weaknesses.

Consider Pakistan’s public sector energy enterprises. Delayed recoveries, rising costs and operational inefficiencies place considerable pressure on their finances. AI-based tools could help directors identify emerging problems, question management and seek corrective measures before financial pressures escalate.

Similarly, audit committees could use AI to examine transactions, identify recurring control weaknesses and monitor unresolved audit observations. This would allow directors to focus less on gathering information and more on evaluating management performance.

Pakistan already has a framework to support such improvements. The State-Owned Enterprises (Governance and Operations) Act, 2023, the SOE Ownership and Management Policy, 2023, and the SOE Audit Committee, Internal Control and Risk Management Regulations, 2024, establish responsibilities for oversight, internal controls and accountability.

The Ministry of Finance’s Central Monitoring Unit also provides an institutional platform for monitoring SOE performance. AI-enabled analysis could strengthen this function without creating another oversight institution.

However, technology cannot compensate for weak governance.

An AI system may flag an unusual transaction, but it cannot establish wrongdoing, compel management to explain it or ensure that directors take corrective action. Nor can it resolve political interference, weak board independence or institutional accountability failures.

Practical challenges also remain. Fragmented information systems, unreliable data, cybersecurity risks and limited technical expertise could undermine implementation. Feeding poor-quality information into sophisticated software will not produce reliable decisions.

A sensible starting point would be pilot projects in selected federal enterprises, supported by clear performance indicators, reliable data, independent validation and board training. Expansion should depend on demonstrated improvements rather than technological enthusiasm.

Ultimately, Pakistan’s public enterprises do not need another sophisticated reporting system that merely documents familiar problems. They need boards capable of recognising risks early, questioning management and ensuring that corrective action follows.

The real test of AI will not be how quickly it identifies problems or how many reports it produces. It will be whether boards use those insights to make timely decisions and hold management accountable.

Technology can strengthen oversight. The responsibility to act must remain human.

Opinion pieces reflect the views of their authors, not necessarily an institutional position of PIVRA.