AI and legitimacy: government in the age of the machine
“They come and take our jobs.”
“They are disconnected from the masses.”
The ‘they’ in the first quotation refers to immigrants and the one in the second refers to technocrats. These two statements, along with other populist ones, helped Donald Trump win the US presidency in 2016.
Replace the ‘they’ in both sentences with ‘machines’, and the slogans capture some of the sentiment against Artificial Intelligence (AI).
Economist Michael Hicks observes that over the last three decades approximately 80% of job losses in America were due to technological advances in the manufacturing sector. Immigration remains the scapegoat to which most of the anger around rising unemployment is directed. The rise of artificial intelligence is projected to bring increases in economic productivity and innovation for the richest countries, but it will also remarkably change the world of work. Some scientists caution that AI may lead to massive unemployment and further polarisation in wages.
There is also a marked resistance to technocrats in government. Whether in 10 Downing St. in London, in the White House, or elsewhere, these elites are accused of being so distant from the lives of everyday citizens that their policies might as well come from a different dimension.
Algorithms are the ultimate technocrats, as they can sometimes be built on maximization models and stripped logic that is perceived as being far removed from the reality of human life. What is more, algorithms are not human. It should come as no surprise, then, that people are fearful of disastrous outcomes if algorithms develop policies, disburse benefits, or determine likelihood to reoffend in criminal procedures
The truth of the effects of AI on societies is likely somewhere between the “technoskeptic” and the “technoptimist” predictions. Regardless of where one falls on the scale, the adoption of artificial intelligence by government raises important questions about government legitimacy.
Trust is the currency of legitimacy. Governments are perceived as legitimate when citizens trust them to make decisions on their behalf. Only when a government is seen as legitimate will citizens consent to the government’s use of coercive power. This trust is build when governments are perceived to be making policies and decisions that are in the best interest of their citizens. Globally, several measures show that trust in governments is eroding, partly because governments have not delivered the sort of evenly distributed economic growth that is desired by their citizenry. It is in the midst of these changing sentiments towards government that public sector applications of artificial intelligence are gaining prominence.
AI will increase productivity in the wider economy, and in the public sector, but these efficiency gains may not have the legitimising benefit that one might expect. This is because governments have to care about more than just the bottom line, and this is a good thing. In a democratic government, it is not enough to tell a jobless person that the economy is growing. Their vote matters, and if they are left behind when automation happens, they may question the legitimacy of their governments, especially if there is a failure to mitigate some of the projected unemployment that could follow AI automation.
American economist David Autor argues that artificial intelligence, like all the technological advancements that preceded it, is unlikely to lead to a reduction in the number of jobs available in the economy. He argues that automation complements labour by raising outputs in ways that lead to higher demand for labour and that interact with adjustments in the supply of labour. Autor is wary, though, of the hollowing out of middle-income jobs and the polarisation of wages that is often part of modernisation.
If the government is to retain legitimacy in the eyes of the populations who are left out, it needs to be preparing approaches now to the social and demographic challenges of a world where AI is part of daily life.