For the last two centuries, workers have been the central factor of production. As a result, real wages in industrialized countries have steadily risen and living standards soared to 20 times what they were 200 years ago. But the Age of Labor—with humans providing the main component of output—may be coming to a rapid end. Artificial Intelligence (AI) could soon reach the level of Artificial General Intelligence (AGI), at which it would surpass human capabilities in virtually every sphere, and governments in both the developed and developing world must prepare themselves for the immense economic, social, and political transformations that this may entail.
It is impossible to predict for sure whether or when AGI will be reached and machines will be able to accomplish the full range of human, labor-related tasks. But the pace of progress indicates that such a seismic transmutation may be closer than many think. In the early 2010s, computers were still struggling with image recognition, the ability to identify objects and faces. Today, cutting-edge AI systems are literally a million times more complex than they were ten years ago, not only superior to humans in image recognition, but capable, as large language models (LLMs) are, of logical reasoning, writing code, doing math, and composing coherent essays. They can also be creative and display theory of mind, the ability to understand human thinking and motivation. Business magnate Elon Musk has tweeted that he would be surprised if machines didn’t have AGI by 2029. Metaculus, a leading prediction engine recently reported that the most likely date for that is 2032. “Mitigating the risk of extinction from A.I. should be a global priority alongside other societal-scale risks, such as pandemics and nuclear war,” warned a group of more than 350 executives, experts and pioneers in the AI field in a statement released May 30, with many of them separately calling for regulation.
Disrupting Labor Markets
What seems clear is that while AI will radically increase productivity, further advances also threaten to disrupt and potentially even destroy labor markets. In the short term (and it is difficult to know exactly what that timeframe means), it may dramatically improve the performance of highly skilled cognitive workers—people like doctors, lawyers, engineers, and academics. It should also provide a short-lived boon to manual laborers who will benefit as economies grow both in the developed and developing worlds. Relatively unskilled cognitive workers—people who work in call centers, for example, or low-level accountants—may be replaced soon, however. And over the longer term, everyone may be dispensable. Increasingly sophisticated AI systems could replace highly skilled cognitive workers as they learn to do the most complex tasks, and manual workers, as they take over robotics and manufacturing processes.
AI and the End of a Development Model
Such a development would almost certainly end the dream of manufacturing-based, export-led growth for developing nations, like the countries of Latin America and the Caribbean. Those countries have looked to the East Asian “miracle” of the last half century as a case study, an example of how developing countries’ comparative advantage in cheap labor can generate a leap forward, leading to an ability to produce higher value-added and more sophisticated products. But that comparative advantage may disappear when AI can engage in manufacturing processes at even less expense.
There will be some significant benefits, at least at first. The owners of AI technology are mostly in the United States and to a lesser extent, the United Kingdom, but advanced systems are being rolled out around the world, and people everywhere now have access to ChatGPT. Provided they develop the proper digital infrastructure, developing countries should be able to continue taking advantage of the latest generations of AI to acquire significantly better education, as well as world class medical, engineering, and other consulting services—without having to import them the traditional way. That, in turn, could lead to large advantages in generating intellectual property in more sophisticated products that could substantially improve developing countries’ terms of trade.
The Economic Fallout of a World Without Work
The biggest concern is what will happen over the long term. Whether it happens in ten or more years, AGI may replace all labor—cognitive and manual, skilled and unskilled. Wages in that phase of the AI revolution would fall for everyone in developed and developing countries and may decline below subsistence level, with mass misery and political instability threatening to upend the social order. At that point, the owners of AI technology in the developed world may cling to monopoly power, ensuring growing inequality and hardship. Or, with some luck, new economic and political institutions will be established to tax their immense wealth and output and distribute the benefits fairly around the world.
Everyone then could benefit from some form of Universal Basic Income, whether that came through owning shares in AI companies or as benefit payments. The Age of Labor would come to an end, replaced by an age without work. People might labor for the satisfaction of it: for the structure, meaning, and social connections they derive from it. They might commit to unsubsidized and voluntary labor, exerting themselves only out of a desire to do so, with exceptions carved out if there are a few areas where human labor is still desirable. Inevitably, for some people, this would be hard psychologically at first, given everything we have long thought about the importance of labor to purpose and self-worth. But eventually with the help of new social customs and institutions, humans could learn to enjoy their free time and flourish in a new era without toil. Such massive transformations may seem beyond imagination, and so are the levels of global collaboration they demand. The emergence of ultra-powerful forms of AI and the epic challenge to global governance it represents is far beyond our experience. But we have created global regimes for dealing with nuclear weapons and nuclear energy; for pathogenic disease; and for global warming, and this is another of our great global challenges. The time to prepare for that moment when AI dominates production and humans create a fair world without work is now.