The 2020s will forever be remembered for the breakthroughs in the world of artificial intelligence (AI). Hundreds of millions of people now use AI in some form, while companies have been pumping billions of dollars into revolutionising their operations through AI. Now a ubiquitous term, AI promises to increase productivity and make the existing workforce efficient. Naturally, Pakistan has also adopted this new technology, but in a style unique to it, its adoption has had so many loopholes. According to a recent survey by cybersecurity firm Kaspersky, a majority of professionals in Pakistan have begun using AI tools at work without receiving any formal training, while most organisations have also failed to provide structured learning opportunities. In a country where the government has announced a national AI policy that focuses on training one million people by 2030, such results expose how fractured policymaking is. In addition to this policy, the government has launched the National Semiconductor Initiative, deeming it key to building a knowledge-based digital economy.
The Inspire programme will train 7,200 professionals over five years in chip design and research across nine universities, supported by the Ministry of IT and the Pakistan Software Export Board. Missing from all these initiatives are well-equipped training centres where both the existing workforce and fresh entrants are taught how to use AI effectively. That so many people are using AI tools without any formal training could make them dangerously dependent on the technology to the point where they will not be able to critically analyse where the technology may have gone wrong. There are cases from all over the world where AI’s reasoning or assessment has been wrong. When we discuss training, we advocate for a learning environment where such examples are cited, allowing people to learn how best to avoid them.
Talks regarding AI adoption are going on at a fascinating pace, but when it comes to action, the government has little to show. At the time of the launch of the AI policy, the government added that it was working on a local LLM, but there is little transparency on how data is being collected to train the LLM – whether we are making a model of our own or using a preexisting one to fine-tune it with local data. All these factors affect our relationship with AI models. Pakistan also lacks the strong infrastructure needed for data processing. High taxes on laptops with GPUs mean most people cannot buy them; this again could create sharp digital divisions, with only the rich having access to the latest technology. The Kaspersky survey should be a wakeup call for our authorities to realise that if we do not act now, we can be left behind. Countries are spending billions on creating local products. In the world of AI, self-sufficiency is important, and the first step towards it is to build a strong internet infrastructure where the internet does not go down every other week and build a workforce that has an excellent command of AI.