About a month ago, my 11 year old son, triumphantly declared that he does not need to learn how to program. "My favourite game 'Roblox' has announced that it is about to launch a Generative AI plugin that will allow anyone to tell the computer to design a game and it will do without the need for any programming."
This, after a few months of me hounding him to learn programming so he could create his own games, rather than just play these. "You must learn to become the creator of new technology rather than just be its consumer," has been my constant retort to him for probably as long as he remembers.
The age of AI
Since late last year, a seismic change has been sweeping through the world of technology with potentially deep and far-reaching consequences for skills and education but, essentially and ultimately, all aspects of our lives in this world. The launch of ChatGPT by OpenAI - a Silicon Valley based non-profit (now converted to a for-profit) dedicated to the development of General Artificial Intelligence (AGI), took the world by a storm and became the fastest growing application to ever cross 1 million active users, doing so in 5 days, defeating the closest previous record held by Instagram by a multiple of 10.
ChatGPT is a simple, but user friendly, interface that allows users to converse with an artificially intelligent system. Based on a technology called Large Language Models (or LLMs), ChatGPT uses large amounts of textual data, primarily drawn from the internet, to train an artificially intelligent system to understand word associations and contexts, and produce an output that resembles intelligent conversation. Well, sometimes!
The underlying technology behind ChatGPT is, perhaps, not new. It has been around for at least a decade, if not more, and language models have been built for many decades now. What has happened, in recent years - OpenAI was only created in 2015 - is the ability of the immense computational power and data to come together to produce models that can deep learn by running calculations and achieve statistical accuracy good enough to mimic human conversation. Two efforts, amongst many in this respect, have been that of OpenAI - founded by Sam Altman the former CEO of Y-Combinator, and funded initially by Elon Musk and others and later by Microsoft - and DeepMind - founded by Dennis Hassabis, and later funded by Google.
While DeepMind has demonstrated some early successes such as machines beating humans in Chess and Go, and protein folding, the last round of this deep rivalry has been won, quite spectacularly, by OpenAI, securing a $30 billion valuation overnight in the process . Microsoft founder Bill Gates has declared that the 'Age of AI' has arrived.
Generative AI's immediate impact on society
The new revolution of Generative AI that ChatGPT has unleashed has the potential to be transformative as well as deeply disruptive for life as we know it. ChatGPT and similar generative AI tools can, in some cases, produce human-like content by searching through vast amounts of data - text, speech, images, and videos - and can generate intelligent media in almost real-time, albeit with considerable, but not prohibitive cost. You can ask generative AI a question, you can ask it to summarize a book or a document, collect and compile data, plan a meeting, or generate an image, an audio, or a video and it can do it with some degree of competence. It is better in some tasks than others, and its latest version, 4.0, is much better in most tasks than its predecessor, 3.5, which is pretty bad and crummy, but these are probably going to be minor details as the full capabilities - including emergent capabilities - of this nascent technology become fully apparent.
The most immediate beneficiaries and affectees of this revolution are going to those with skills and jobs that can be significantly eased or enhanced by generative AI chatbots and computers, but also replaced by them. The more predictable and mundane the task is, the easier it would be to be fully or partially replaced by these tools.
OpenAI, the creator of ChatGPT, estimates that approximately 80% of the U.S. workforce could have at least 10% of their work tasks affected by the introduction of GPTs, while around 19% of workers may see at least 50% of their tasks impacted. "The influence spans all wage levels, with higher-income jobs potentially facing greater exposure." Suddenly all these years of predictions by the likes of the World Economic Forum (WEF) and others of massive disruption to global employment are no more predictions in some distant future, but very much real and knocking at our door.
Potential impact on jobs, skills, and lives in Pakistan
What is even more challenging, however, is the real likelihood that countries like Pakistan may find themselves at the punishing end of this disruption. It may very well be that amongst the educated Pakistani middle class, at least, the above ratio could be reversed, with 19% of the educated middle class workers seeing 10% of their work tasks affected and as many as 80% of them seeing 50% of their work tasks affected by AI. Anybody whose job does not require physical work - at least for now - is at the risk of getting replaced - partially or fully. Those performing average secretarial and knowledge work - such as research, writing, content, accounting, designing, teaching, training, planning and even coding - are likely to face the brunt of the generative AI revolution. Ordinary content designers, writers, researchers, and software developers beware. The 'one day' when AI impacts 'our jobs' has arrived.
By and large, ChatGPT, MidJourney and tons of other generative AI applications can do in 15 minutes what an ordinary writer, researcher, or content developer in Pakistan would usually do in a whole day, and they can do it much better - with perfect English and fewer errors. Give it the right prompts and it will churn out a document, a plan, and design, or a piece of code. No need for the army of people toiling away all day to try to understand your instructions and implement them for you. Yes, these employees could perhaps do their jobs better if they also used Generative AI but so could the supervisor. The former will require a nontrivial amount of retraining and reimagining of knowledge-based work.
Many of these switchovers will happen almost immediately and their impact will be obvious within a few months - not years - as we see the demand for knowledge workers declining in the face of perhaps the worst economic crisis Pakistan has seen in its life. Many will blame the economics for this decline but many of these jobs may never return back even as Pakistan goes through an economic recovery primarily because most of these 'low performing' knowledge workers will not be needed anymore.
Unfortunately, as a country, we have never really paid proper emphasis on creating knowledge workers - Knowledge Economy Task Forces and other slogans aside - as knowledge work requires a strong foundation of thinking, communication, and collaboration skills, which are most effectively imparted in quality undergraduate programs.
These deep structural issues aside, there is still, perhaps, some redemption for the Pakistani middle class workforce as ChatGPT and related technologies also promises to paper over some rough edges - linguistic and productivity challenges of our workforce - and create an enabling environment for those who can leverage these technologies to get ahead. But this will require smart thinking and smart work on the part of many of us.
Thus, in this deeply disruptive crisis, perhaps, there is also a glimmer of hope and an opportunity to transform our education systems, our skills, our lives, and our destinies.
The way forward
Back to where I started from. My rejoinder to my 11-year old son who thought he doesn't need to 'learn to code' in order to create technology went as follows:
There will be four kinds of people in the world. The first will be those who who build transformative technologies and tools like ChatGPT and others. They will need to learn how to code. The successful ones among these will probably end up making hundreds of billions of dollars. Then there will be those who will use the tools, such as ChatGPT to build businesses on top of them. They may not need to learn how to code, but will certainly need a lot of very important and diverse business and people skills. The successful ones amongst these will end up making, perhaps, 10s of billions of dollars. Then there will be thos ewho will use the tools, such as ChatGPT and others, and be able to integrate, synthesize, and interpret knowledge, code, and media generated by it. They will end up becoming technical and managerial leaders and make, perhaps, 10s of millions of dollars.
Finally, there will be those who will either become ordinary users of these tools, and hence, will barely be able to save their jobs or will refuse to learn to adapt to the new reality and end up losing their jobs altogether. They will live hand-to-mouth, survive on basic incomes, or fail to provide for their families.
The world, over the next 20-50 years, will see even starker disparities between the digital haves and have-nots.
Regardless of our misgivings about AI, countries like Pakistan have no choice but to embrace this new revolution, which is likely to be at least as disruptive if not more than many of the previous ones - the silicon chip, computers, and the internet.
We must train our workforce to become intelligent users and integrators if not the creators of the new technologies in the age of AI.