Upcoming AI policy may turn out to be an exercise in futility

The paper says the Pakistan government's upcoming AI policy has been crafted without finalising dedicated data laws, which have been in the draft stage since 2018.

ai.jpg

May 30, 2023

ISLAMABAD – Ever since the launch of ChatGPT, you must have come across countless takes on the potential or consequences of artificial intelligence (AI) — from doomsday peddlers to grifters trying to cash in on the hype with their meaningless tips. Somewhere in between, there is a genuine concern among professionals and policymakers on the technology’s potential to unravel and human inability to prevent that.

In this regard, the United States Senate recently held hearings where Sam Altman, the CEO of OpenAI — the organisation behind ChatGPT — remarked: “I think if this technology goes wrong, it can go quite wrong”.

He’s among the many leading voices, including Elon Musk, calling for regulating AI. Obviously, that’s easier said than done since no one really knows how to really go about it, at least the enforcing part.

But that doesn’t stop policymakers from coming up with detailed plans possibly governing this technology. The Pakistani government, too, jumped on the bandwagon recently when the Ministry of Information Technology and Telecommunication put out a draft National AI Policy.

God knows how many trees had to give up their lives to make the paper for a 41-page framework that might never see the light of the day, at least in spirit.

The upcoming AI policy has been crafted without finalising dedicated data laws, which have been in the draft stage since 2018

What makes this whole exercise hilarious is that we have an upcoming AI policy without finalising dedicated data laws. The Personal Data Protection Bill has been in the draft stage since 2018. During this period, Pakistan’s economy has seen three cycles, the exchange rate has worsened from less than Rs116 to Rs284, and the establishment’s control over politics is finally over…oh wait.

Yet, once again, despite the foresight that the National AI policy is most likely nothing but a waste of paper and some Islamabad-based consultant’s extra cash for a summer holiday, let’s look at what the big plan really is, if there is really one.

The draft is based on four pillars: 1) Enabling AI through awareness & readiness, 2) AI market enablement, 3) Building a progressive and trusted environment and 4) Transformation and evolution — further categorised into 15 targets.

The first pillar starts with proposing the establishment of a National Artificial Intelligence Fund, for which the Research and Development Fund (Ignite) will allocate at least 30 per cent of its money on a perpetual basis. Of course, to administer it, you’d need up to 11 board of directors and a chief executive officer. Good times are here to stay (for boomer uncles).

Staying true to form, it then proposes the most pressing requirement for the uptake of AI in the country: a network of Centres of Excellence. I mean, the construction industry needs to make a living too. It’s already been tough for them without the amnesty scheme and the cuts in the Public Sector Development Programme. Sure, they are to be more than just buildings, you know, like those incubators turned out.

The last target of this pillar is valuable, for it proposes “standardising the data of state-owned enterprises, boards and civil authorities for applying AI-based algorithms”. This will be then made available for processing through a sandbox-based licensing approach under the Centres of Excellence. Think of large health, education, and utilities datasets opened to tech companies to train their models on.

The document delves into potential use cases for each sector, from transforming health services to intelligent learning and assessment. However, this whole section reads a lot like someone asked ChatGPT for AI applications. Sample this part: “The biggest challenge in the agricultural sector regarding supply chain optimisation is data digitisation and standardisation.” I am no agri expert but wasn’t our biggest challenge the lack of infrastructure? You know, like cold storage and efficient transportation?

Moreover, the policy lays out lofty targets for training the workforce — with a target of upskilling one million existing and new IT graduates in “high-impact applied skill in AI and Allied Technologies” by 2027. For this, 10,000 new trainers will be needed. Through the centres, the ministry also plans to organise an internship program which will offer at least 20,000 placements annually in collaboration with the private sector.

It also envisions a scholarship programme which will fund 3,000 deserving candidates on open merit every year in postgraduate and doctoral programmes across the country. Additionally, there’d be 500 seats for women and 100 for persons with disabilities.

The policy seeks to encourage the production of academic knowledge on AI from Pakistan and, through Centres of Excellence, proposes logistical/fiscal/data support for up to Rs200,000 per thesis or paper, leading to publications in regional or international journals.

An AI directorate will also be set up with the goal “to augment the development of AI-based initiatives through need-based assessment of regulating different functions automated using AI and the type of data being created and processed for defined purposes.”

For the sake of brevity, a lot of details from the policies have been left out, specifically those relating to the modalities and processes. This is what the whole thing really seems to be about: ticking checkboxes. AI is all the hype these days and countries around the world have produced their strategies for dealing with it, so how can we stay behind? Even if the paper it was printed on will be used to serve samosas.

scroll to top