Governments may approve artificial intelligence taking human jobs where profits soar and can be shared with workers, new research has revealed
Democratically elected governments may willingly let artificial intelligence replace workers – and risk higher levels of unemployment – if the technology shift leads to soaring profits that can be redistributed to those affected, according to new research from Durham University Business School and the University of Bath.
Whilst the researchers state that the advent of automation is potentially good news for worker welfare in industrialised democracies when automation is highly productive and can relieve worker pressure, they warn it is a far less promising prospect for workers in otherwise populist governments, which may allow the economy to over-, or under-invest in automation, damaging worker welfare and prospects.
The study, conducted by Dr Bibhas Saha, Associate Professor of Economics at Durham, alongside, Jaideep Roy, Professor in the Economics Department at the University of Bath, aimed to understand the conditions under which governments would support an aggressive rollout of AI in various sectors – even where it may lead to job losses and possible voter backlash – versus the situations where they choose to slow down AI adoption to protect employment and maintain public support.
Looking for the right balance
The researchers are the first to develop a novel theoretical model that examined the trade-offs between efficiency gains, unemployment risk and political costs. This allowed them to explore how varying levels of AI-driven profits could influence electoral outcomes when automation of the workplace became an important element of the election agenda.
The researchers identified a clear conflict between governments’ desire to invest in AI to enhance efficiency and their need to protect jobs and avoid mass unemployment. However, they also found that a balance can often be struck between investing in AI for efficiency, limiting unemployment, and maintaining voter support.
Two types of political equilibrium can occur. In one, profits generated from the high scale of automation are substantial enough to meaningfully compensate all workers affected by automation implementation. In the other, automation is restricted and AI-driven profits are low. As a result, these are shared only with protected workers, whilst the affected workers go uncompensated. This second possibility, the researchers warn, is concerning.
“As countries like the UK invest in AI to drive growth, they must carefully balance efficiency gains with the risk of mass unemployment,” says Professor Saha. “Thoughtful regulation is crucial to ensure that automation benefits society while protecting workers from being left behind. Without the right regulatory framework, the rapid adoption of AI could exacerbate inequality and disrupt the very workforce it aims to improve.”
Considering novel approaches
Given that investing in AI is a government priority with significant economic benefits, the researchers state that regulation to protect workers is vitally important. They suggest that governments should consider novel approaches such as robot taxation, enhanced social benefits, or workforce transition support to ensure workers are not disadvantaged by rapid AI implementation.
Further reading
The study, Democratic regulation of AI in the Workplace has been published by Elsevier in the Games and Economic Behavior Journal.