WHEN asked what he would dress up as for Halloween, a 10-year-old boy I know responded, “Claude, because there’s nothing scarier than AI.” Whether you consider this funny or astute, it highlights the pervasiveness of the AI bogeyman. And the best thing to do with your fears is face them.
So it was good to see Pakistan play a prominent role at the recent Global Digital and AI Summit in Seoul and ramp up engagement with the World Bank to support the country’s digital transformation through a Digital and AI Compact. Pakistan has committed to spending $1 billion on AI by 2030 and is cracking on with plans to digitise 12 key sectors by 2035 under the National Digital Master Plan.
As the momentum builds, we must ask: how will Pakistan fund investment in AI? And what makes digital transformation and AI adoption capable of overcoming structural issues that have held back traditional activities like manufacturing?
AI use — particularly localised AI development and large language model (LLM) training — requires an immense and stable energy supply, high-speed internet and data storage, none of which are readily available in Pakistan. While efforts are underway to boost 5G spectrum and satellite access, the electricity sector is plagued by outages, a dilapidated grid, and circular debt. Concerns abound about whether Pakistan’s scarce water supply can bear the burden of water-guzzling data centres.
Then there’s the question of how to insulate the future of Pakistani youth from widespread AI adoption. Pakistan’s first economic census last year identified that the services sector makes up over 50 per cent of the country’s economy (production is less than 30pc).
How will Pakistan fund investment into AI?
A recent World Bank report notes that the services sector is most vulnerable to AI adoption, with 61pc of Pakistani service workers in jobs with higher-than-average AI exposure. Based on 2024 data, the report indicates that 8pc of all jobs are at risk of being replaced by AI (a number likely to rise). This means many Pakistanis will be trapped in low-skill, low-paying, and climate-vulnerable agricultural or manual labour jobs, while a sliver of society benefits from productivity gains.
The other question is how to prevent the complete securitisation of the AI agenda, a common challenge in countries with national security preoccupations. Concerns about the use of LLMs for mass surveillance have been voiced in the US and China, a scenario that Pakistan, with its punitive electronic crime legislation, is well primed for. Early interest has also been expressed in using AI for smart policing and biometric identity management. An early, advanced data centre was built under a National Aerospace Science and Technology Park initiative.
The state will feel a growing need to direct AI investment towards national security, because it’s only a matter of time before local terrorist groups begin deploying AI in anti-state activities. Writing in South Asian Voices, Zohaib Altaf and Nimrah Javed highlight the risk of such groups using AI to better target attacks based on intelligence analysis or to support arms development, particularly weaponised drones. The precedent for this has already been set by the Houthis in Yemen, so the likelihood of AI investment being funnelled primarily to military ends is high.
If Pakistan’s AI agenda is not carefully monitored, we face a future in which limited digital infrastructure in the form of resource-intensive data centres is reserved for security-linked AI use, with some private enterprises enjoying gains at the expense of the wider population. In this scenario, urban youth unemployment increases and rural poverty intensifies; agricultural employment survives the AI rollout, but is ravaged by climate change instead.
There is a way to prevent this scenario. That’s through high levels of transparency and accountability in AI governance, regulation, and investment, including the requirement for public service considerations and consultation as part of all investments. Pakistan’s civil society organisations and media should also redouble efforts to understand the implications of AI investment to ensure accountability.
On a more positive note, the notions of accountability and transparency in AI are no longer antithetical to AI development. US-based civic activism against data centres on resource-use grounds is increasingly effective in blocking projects. And while Big Tech has thrown up its hands, arguing that it can’t be held responsible for what LLMs do, new research suggests that LLMs may be easier to unpick (and so monitor) than we think. Such public-interest considerations, particularly with an inclusion lens, should be part of Pakistan’s AI journey. That may make it less scary.
The writer is a political and integrity risk analyst.
X: @humayusuf
Published in Dawn, October 26th, 2026
