KARACHI: Amid efforts at regulating AI, the government of Pakistan may be missing out on an important opportunity: tapping into the cost-saving potential that AI offers. The News reached out to experts and asked how the state can save on its expenditure by adopting AI to eliminate manual work in areas that need less human intervention to ideally allow people to focus on skill development and to reduce spending.
Chief Executive Officer of the Pakistan Business Council (PBC) Ehsan Malik says: “That AI is going to change how we work is a certainty. For a country like Pakistan to fully benefit from AI, we need a higher level of digitization together with affordable, high speed and widely available connectivity. At the same time, we need to simplify and make government decision-making transparent. A regulatory environment run by a bloated bureaucracy with a colonial-era ‘control and extract’ mindset needs radical change. This bureaucratic drag requires strong political will to overcome.”
He adds, “industry, however, can move faster to adopt AI in their processes. Our IT sector is already addressing the global AI-led software demand.”The private sector has taken good steps to tap into the global IT sector. The country’s IT industry association, PSHA, continues to play a pivotal role in bringing tech companies to the global stage. In August, ITCN Asia held the second edition of its tech exhibition in Karachi (the first one was held in Lahore) to allow international tech companies to meet Pakistan’s tech talent.
But the government’s adoption of AI has been rather disappointing. Umair Arif, who has a PhD in artificial intelligence (AI), offers various ways through which the government could make the most of AI technologies. He explains that AI has three main branches: vision AI, predictive AI and language AI. “The government can employ vision AI to keep a check on traffic violations and improve vehicular traffic. Language AI can be used for writing-related work, so that a large team of people is not required to write proposals or other bureaucratic work. Similarly, it could use predictive models for budget preparations and financial predictions to see what problems it could face in the future.”
On the face of it, the government does recognize the potential of AI. In Sindh, the provincial government recently launched the People’s Information Technology Programme where it partnered with engineering universities in the province to provide free IT certificates to college students. Similarly, in her press conference last Friday, Minister of State for Information Technology and Telecommunication Shaza Fatima Khawaja said that Big Tech’s interest in Pakistan (referring to investments made by Alphabet’s Google) was a sign of the country’s IT potential.
While all of these are good indicators, Pakistan still has a long way to go to leverage the revolutionary technology. Education expert Dr Ayesha Razzaque explains why such adoption is a little tricky: “A lot of recent AI applications seen around the world have been enabled by large language models, or LLMs, many of which do not provide a lot of support for our local languages whereas a lot of paperwork in Pakistan still gets done in Urdu and regional languages (think FIRs and official forms).”
She adds that the development of language models domestic to our country is an expensive undertaking. A few private companies have made some attempts in this direction but lack the material resources that are needed for a serious effort.
“For the moment, that only leaves pre-trained models many of which are available for subscription.” But these tools come with their own set of challenges. Ayesha says, “If you hope to improve the efficiency of tasks that can be performed in English, you still need a human being that is capable of verifying the correctness of the output such tools produce.”
“If you hope to apply AI models to existing organizational data, that presupposes that data is available in electronic form and does not exist in silos. As of today, as per my understanding, neither of these two conditions have been fulfilled.”
Red tape in Pakistan has long been an issue for industries in different sectors. Taking a jab at the government’s infamous slow speed of doing things, Ehsan says: “AI emulates human behaviours but does it much faster. If AI ‘learned’ from the way bureaucracy has behaved for 70 plus years in Pakistan, it would perpetuate slow decision-making.”