This essay was written for The Economist’s ‘Open Future’ writing competition and was shortlisted.
When the histories of our age are written, Donald Trump may be nothing more than a footnote for political scientists and sociologists (if not psychologists). Instead, our era may well come to be defined by our development of early artificial intelligence, and assessed on how we respond to that ultimate Pandora’s Box. The reason is simple: AI presents opportunities and challenges on a scale comparable only to the Industrial Revolution or the Agricultural Revolution. The progress of driverless cars or DeepMind’s Alpha Go are important stories in their own right, but discerning readers would be well-advised to also consider the wider picture.
Consider, for instance, framing the question of artificial intelligence in the context of inter-country inequality. There are a variety of reasons why developing countries are likely to struggle rather than thrive as the AI Revolution gets properly underway. The first is the rapid erosion of the developing world’s traditional labour cost advantage. Notwithstanding just how low the cost of labour is in many developing countries, AI-powered robots and machinery are already at the tipping point of cost efficiency compared to human workers. As technology continues to improve, the sheer economics of AI will see a whole swathe of companies replace their factories and operations in the developing world with new AI-powered facilities staffed with robots. This wave of ‘reshoring’ (the rolling back of ‘offshored’ work) has been flagged even by the World Bank in their 2019 World Development Report.
This decimation of manufacturing jobs in the developing world will be particularly harsh for states reliant on ongoing investment from foreign corporates in manufacturing, like Vietnam and Thailand. Not only do those countries rely more heavily on sectors like manufacturing that are most prone to automation and disruption, but job losses within those sectors are disproportionately felt by developing countries. When Nike eventually decides to automate its manufacturing processes, heads are hardly likely to roll in its plush headquarters in Washington County. Rather, the hundred thousand men and women working in Nike factories in Indonesia will be left with bereft prospects.
Neither is the damage likely to stop at manufacturing. Cheap, ubiquitous robots combined with intelligent AI systems will disrupt industries of all kinds, but particularly those reliant on manual labour or repetitive work like agriculture and energy. Again, developing economies are disproportionately reliant on those industries at the most immediate risk of large-scale redundancies brought about by automation.
Most concerning of all is how keenly the disruption and job losses will be felt in combination by developing countries. Simply put, Thailand and Indonesia do not have the tools or levers to deal with mass-scale automated redundancies. Rich states like New Zealand have both the flexibility to proactively plan for the retraining of people made redundant by AI, and the wealth to provide a safety net for those who slip through the cracks. For a variety of reasons, developing states are much less likely to engage in this necessary forward planning. Generally speaking, developing states have weak governance structures and institutions, often encumbered further by corruption and kleptocracy. Developing states also tend to have more immediate issues like famine, poverty, terrorism or civil war to deal with.
Consequently, the AI Revolution is likely to inflict the most damage on the exact countries that are the least able to respond proactively or effectively. This places individuals and families in the developing world in a precarious situation. They are likely to be younger, poorer and less educated relative to their equivalents in developed states. They will have no choice but to try to get a foothold in an economy assailed by AI, as job openings in agriculture, manufacturing and energy evaporate. It is difficult to see how these men and women adjust to the new paradigm. The research available suggests that adjusting to a new career in the wake of automation requires resources and re-education, and often government support. These are bleak findings for a factory worker or farmer in Southeast Asia.
To be clear, I do not mean to forecast some sort of permanent economic Armageddon for the developing world. The masses of unemployed will surely include an outstanding selection of entrepreneurs and innovators, many of whom will harness AI and create new businesses and plenty of jobs. Otherwise, the informal rural economies of many developing countries may prove resistant to market forces. Deploying AI or robotics in the fragmented Indian farming economy is no easy task. However, it seems inescapable that the greatest costs and risks of the AI Revolution will be inflicted on the developing world. The benefits of AI must accordingly be weighed against its impact on the gulf between developing and developed countries. While the existence of such a gulf is nothing new, the introduction of AI introduces some new and dangerous possibilities to the equation.
Divide and Conquer
The first such possibility is the likely growth of a Luddite political class looking to assuage the concerns of the suddenly jobless, who seek a defence of their livelihood and their family’s future. No matter whether the state is controlled by a democratic government or a military junta, the most powerful political currency in the developing world may soon be the anger of a large number of unemployed people. The populists abreast this wave of anti-technology fury will be a danger both to their countries and to world order. You can safely count on a slew of nonsense economic policies and ineffective regulations seeking to hold the force of technological progress at bay. More dangerously, these populists are very likely to lump the blame for the unemployed’s woes at the feet of those countries reaping the benefits of artificial intelligence: the developed world.
Relatedly, the second threat of AI is the entrenchment of the existing gulf between developed and developing countries. The AI Revolution will undoubtedly serve as a skyrocket for the developed world: rich countries and large corporates are already at the forefront of the technological surge and are well placed to secure the gains to follow. If AI serves instead as a lead weight for developing countries, like we have at least some cause to fear, prospects for closing the gap between rich and poor countries are bleak. It is well documented that the economic rise of China is the chief cause of millions of Chinese people attaining better jobs and lifting out of poverty. AI may prevent a similar rise for countries like India and Brazil, by permanently concentrating economic power in the developed countries that control advanced technology.
Similarly, AI risks a long-term handicapping of developing states in the context of future advances in technology. Even if job losses and disruption by AI does not cripple developing states and their economies entirely, it seems inevitable that those states will face a torrid period of short and medium-term adjustment to a new post-AI economy. That adjustment will suck up significant resources and have an outsized effect on the politics of that society. Consequently, they are hardly well-positioned to invest in research and development, further learning, or otherwise prepare for the next technological breakthrough. Of course, that means developed nations are likely to have an even stronger control of gene-editing, or cybernetics, or whatever the next technological revolution is. The risk here is that AI will pose such a disruption to developing states that the bounty of cascading technological advances is concentrated elsewhere (yet again).
At this stage, however, these dark scenarios remain hypothetical. The histories of our time have not been written yet. But what seems plausible, if not downright likely, is that developing countries will soon face one of the most difficult economic challenges imaginable. How will states react if 10, 20, 50% of their working population become redundant over a decade? What happens to the relative stability of the global order? We had better start thinking about answers, and quickly.