Why is no one talking of AI in Pakistan?

Azher Sultan
5 min readMay 19, 2018

--

Artificial Intelligence, AI: the ability to build intelligent machines. Machines capable of driving a car, holding a conversation, beating humans in the most complex board games and identifying a fugitive from a crowd of 60000 people. The world is currently both awed and horrified by its potentials. USA, China, Russia and Europe have invested trillions of dollars so far in AI research, hoping to be the world leaders in it. As President Putin puts it, the nation that leads in AI will rule the world.

In the same world exist countries like Pakistan (and probably many other developing countries). Peace and quiet, as far as the AI is concerned. You can spend months here and not a whisper of it will be heard. Government, media and even most of the universities seem to be completely ignorant of the revolution that is hitting the wold. AI driven autonomous cars are a very foreign concept here and while people do see videos of google assistant booking appointments, technology behind it is rarely read upon.

This would not be a big deal considering the fact that new technology usually arrives in developing nations a decade later, except that AI is the biggest opportunity for a developing nation to jump ahead. With meager resources in a country like Pakistan, it is very hard to compete with the outside world. Infrastructure development is an uphill battle and industry tends to struggle in the absence of right infrastructure. Add the political instability, corruption, international image to the mix and you will be looking at a very grim picture. This is validated by numbers with ~$40 billion spent on imports and and ~ $20 billion gained via exports. This deficit of ~$20 billion is mostly recovered through remittances sent by overseas Pakistanis.

It doesn’t have to be like this. The AI world is facing an immense shortage of skilled labor and annotated data. UK is facing a shortage of AI experts where demand exceeds supply by 2 to 1. Same is the case world over. Most of the experienced professionals are lured by big corporations through salaries and perks. This leaves startups and small businesses scrambling for anyone with some experience in AI.

Considering the billions of dollars being poured into AI startups, this demand for AI expertise is not going to subside any time soon. Following the demand-supply trend of the market, the salaries for AI experts are starting to skyrocket. Startups and smaller businesses are either forced to pay such high salaries from their limited resources or they go the freelance way, to seek out AI experts worldwide at cheaper rates. Hence, AI expertise are one of the hottest in demand skills at freelance websites like upwork.com. This is a great opportunity for developing countries to train the relevant skill force and earn their piece out of the AI pie.

Getting to the minimum requirements of an AI expert is not that hard if you have some knowledge of computer programming. There are a lot of online resources to help you learn. Fast ai and machine learning by andrew ng are two of the many online courses that can get you quick insights into theoretical and practical concepts. Participation in a couple of Kaggle competitions can give your profile the right boost. All in all, it only takes few months to an year to get the prerequisites to be hired as an AI engineer and you do not need any degree for it.

Pakistan, on the other hand,has an abundance of skilled but jobless engineers and other graduates. An engineer with a good job earns about Rs. 85k a month (~$4/hour) whereas on average the salary is much less. In the developed world, an AI expert costs roughly $40/hour (calculated from average yearly incomes with conservative estimate). This is a 10-folds difference in earnings. Coupled with the imbalance in demand and supply, our workforce once trained properly can fill this gap and open another avenue to progress.

Another big need of the AI industry is annotated data. Data annotation means providing examples of solved tasks that AI software can use to learn from. For example, if you are building an AI tool to recognize cats from images, you will first need to manually indicate where the cat is in a huge amount of images. This is a tedious and time consuming task that requires no skills other than the ability to understand basic instructions. Currently companies are willing to outsource these data annotation tasks for $4–10 per hour and these projects can span months to years. Considering the minimum wage in Pakistan, a company established in Pakistan can offer a pricing as low as $1.5 per hour and still be profitable.

Interestingly everyone has a role to play in making Pakistan the hub of AI workforce: Government needs to provide support and legal framework that makes Pakistani workforce lucrative for international players. Universities need to start responding to the needs of the market. AI expertise should be treated as vocational training and institutions across the country should train individuals with minimum entrance requirements. Youth should get some experience into AI no matter what field they belong to, because it will spread to every field in a matter of few years and parents should push their kids (if they are willing) into computer science. If nothing else, help establish institutions that can train the less fortunates of our society for AI.

Pakistan already ranks among top 5 countries in the freelancing circles. Software exports have risen to $2.9 billion in 2017 and there is a huge potential for growth. Punjab government has started a commendable e-rozgar scheme to train people for different freelance fields. Unfortunately it does not currently include AI related expertise, but it is a step in the right direction. AI is a gold rush and while we don’t have the resources to compete in the race, we can still get our chunk out of the competing parties. We had once said that we will eat grass but get an atom bomb of our own. Well, AI is the next bomb and we don’t need to eat grass for it.

--

--

Azher Sultan

Data scientist/AI expert with interests in experimentation and education