Reducing AI risks to human life

THE 2012 book Why Nations Fail: The Origins of Power, Prosperity, and Poverty, by Daron Acemoglu and James A. Robinson, reshaped the way we understand the trajectories of economies, human societies and civilization as a whole. Citing examples from ancient Rome to the 21st century, the authors suggested criteria common to successful societies throughout history.

Casting aside previous explanations based on cultural, religious, geographical, and ecological factors, Why Nations Fail focused on the creation and preservation of ‘inclusive’ versus ‘extractive’ institutions as a key driver of democracy and, by extension, economic success, political stability, and human rights.

The authors argued that one by-product of inclusive institutions was a tolerance for ‘creative destruction,’ the process by which antiquated technologies, laws, or institutions become destroyed as they are replaced by new ones that are more effective at serving a changing society.

Railroading society

To do this, the book cites an interesting historical example contrasting the development of railroads in Britain versus the Hapsburg Empire through the 19th century. The authors argued that in the Hapsburg Empire, elites controlled the roads and charged high tolls, such a system was an example of an ‘extractive institution’ maintained to enrich a small group of individuals at the expense of the larger society. Consequently, it also meant that the elites had little incentive to allow this system to be up-ended by railroads. Conversely, Britain’s slightly more inclusive political and economic institutions permitted changes in the power structure that allowed a new set of individuals to supplant existing systems with superior technology.

Power balance

Can we use this framework to shine some light on questions about integrating artificial intelligence into human life? One could make the argument that the creative destruction framework laid out by the authors would imply that societies with inclusive institutions should be able to tolerate changes to existing power structures that will come alongside the integration of any technology, including AI. Societies that do this will also become more efficient, standards of living will increase, and people will be happier.

However, given the wide-ranging economic applications for AI, the creative destruction framework seems to fall short of explaining the full scope and scale of the changes that will take place if or when AI is integrated into our economies. After all, intelligence is the most defining characteristic of humanity. The authors used the creative destruction framework to understand rapid changes to specific sectors of the economy such as transportation or industrialization.

Move over, Human

However, it’s becoming clear that supplanting humans with a more superior AI will have immediate implications for nearly every form of economic activity. Such a drastic change in the economic order will almost certainly produce drastic political changes, which will lead to instability and lower productivity and standards of living.

In response to this, politicians such as Andrew Yang in the US have proposed a ‘technology tax’ to support workers that have been replaced by machines. Yang and others have also been arguing that tech tax revenues can go toward universal basic income or UBI given the scale of these job losses. Perhaps bold plans such as a technology tax, UBI, or similar measures, represent a move to make our political and economic institutions more ‘inclusive.’ The creation of more inclusive institutions might be required to weather the changes, reap the economic benefits, and reduce the risks of advancing toward a new economic order.