Page 84 - CMA Journal (July-August 2025)

O TH ER F EATURES

Government AI Readiness Index 2024 reliability, and siting challenges. Converting
(South & Central Asia) allocations into cost-effective, low-latency compute
is complex.
v
Government AI Readiness Index 22024 Scoree • Budget Execution & Pipeline Readiness – Flagship
e
A
70 62.81 AI initiatives often appear as “unapproved” or
60
50 38.78 40.47 445.29 47.122 provisional in federal plans. PC-I preparation,
1
40 31.43 33.14
30 16.92 approvals, and procurement capacity will determine
20
10 implementation pace.
0
• Public-Interest Safeguards – Civil-society reviews
flag risks including surveillance, algorithmic bias,

Source: Government AI Readiness Index 2024 due-process gaps, and the gender digital divide,
which require rules, audits, and redress mechanisms.
The Government AI Readiness Index 2024 shows • Standards, Safety, and Ethics – Gaps in data quality
significant variation across South Asia, with India (62.81) and availability, as well as ethical oversight, highlight
leading the region, followed by Bangladesh (47.12) and Sri the need for ISO/IEC, OECD/UNESCO standards,
Lanka (45.29) in the mid-range. Pakistan (40.47) ranks testing labs, and regulatory competence.
below these peers but ahead of Bhutan (38.78), Nepal • Skills & Institutional Readiness – Ministries and
(33.14), and the Maldives (31.43), while Afghanistan (16.92) regulators (PEMRA, PTA, HEC, FBR) need funded
remains the least prepared. Since higher scores indicate positions, clear job descriptions, and comprehensive
better readiness, Pakistan’s position in the lower-middle training programs beyond pilots to ensure effective
tier highlights the need to strengthen institutional and AI adoption.
technological capacity. However, the recent launch of
Pakistan’s first National AI Policy in 2025 marks an • Legal/Regulatory Vacuum in High-Risk Uses – AI
important step forward, aiming to improve governance, applications in justice, policing, and biometric
expand digital infrastructure, and enhance AI skills systems are advancing faster than sector-specific
development—measures that, if effectively implemented, safeguards, creating litigation and rights risks unless
could boost Pakistan’s readiness in future assessments. NAIP-2025 is complemented with sectoral rules.
The policy seeks to strengthen digital infrastructure with
Global AI Benchmarks: MMLU-Pro 2,000 MW of surplus electricity for AI data centers and
(Overall Accuracy) initiatives such as the National AI Fund, Centers of
Excellence, and large-scale skills development. However,
MMLU-Pro: Overall Accuracy (inn %)
c
the absence of a comprehensive data protection law and
restrictions on cross-border data flows remain key hurdles
84.0%
80.3% for effective implementation and global integration.
77.6% 77.9% 78.00%
%
75.5% 75.7% 7 75.9% 76.2%
Conclusion
%
Pakistan’s first AI policy provides a strong foundation with
defined targets, governance frameworks, and ethical
guardrails. Supported by growing IT exports, expanding
Source: Artificial Intelligence Index Report 2025 connectivity, and youth engagement, the country is
These benchmarks demonstrate the rapid advancement well-positioned to elevate its AI ecosystem. The critical
of frontier AI models globally, with accuracy now next step lies in execution—resourcing national
exceeding 80% in complex reasoning tasks. This institutions, setting measurable milestones, and building
highlights the urgent need for Pakistan to invest in local accountability structures. If implemented effectively, this
AI innovation and skill development to reduce reliance policy could transform Pakistan’s digital ambitions into
on foreign technologies. inclusive, AI-driven national progress.
National AI Policy 2025: Challenges and References
https://www.arabnews.com/node/2611685/pakistan
Implementation https://moitt.gov.pk/SiteImage/Misc/files/National%20AI%20Policy%20Consultatio
n%20Draft%20V1.pdf
• Data Protection & Cross-Border Flows – Pakistan https://datareportal.com/reports/digital-2025-pakistan
lacks a comprehensive, GDPR-style data law. Drafts https://www.brecorder.com/news/40332283/ai-could-add-10-20bn-to-pakistans-gd
p-by-2030-says-saudi-data-and-ai-authority-advisor
foresee heavy localization and a Data Protection https://www.thefridaytimes.com/23-Jul-2025/pakistan-s-youth-crisis-deepens-as-jo
Commission, both critical for AI datasets, model bs-education-and-hope-run-dry
https://hai.stanford.edu/assets/files/hai_ai_index_report_2025.pdf
training, and government enterprise cloud
https://oxfordinsights.com/ai-readiness/ai-readiness-index/?#download-reports
initiatives.
This article is prepared by Maiyra Ahmed, Assistant Director, R&P,
• Energy & Compute Economics – Plans to power AI under the guidance of Shahid Anwar, Director R&P)
data centers with “surplus” electricity face tariff,

82 ICMA’s Chartered Management Accountant, Jul-Aug 2025