Skip to navigation

Skip to main content


The Future of Work

July 18, 2019

The world of work is changing rapidly, as machines replace humans in ever more jobs, the traditional employer-employee relationship morphs into workers become independent contractors or having only short-term contracts and part-time jobs, and economic inequality widens socio-economic divisions in many societies.

The starting point for many discussions of the future of work is technology, especially the use of artificial intelligence to make machines smarter. AI can be considered weak or strong. Weak AI performs relatively narrow human-like tasks repeatedly, such as interpreting X-rays or helping drivers to avoid collisions. Strong AI learns over time, as with spam filters that adjust their behavior.

Strong artificial intelligence is evolving to help machines make ever more human-like adaptations. Machine learning algorithms permit computers to improve their performance over time, as with customer service bots. The accumulation of data can lead to deep learning that enable machines to use neural networks to filter and learn from large datasets and generate probabilistic responses, as with speech recognition and language translation.

Artificial intelligence is poised to disrupt labor markets, forcing many workers to change employers and occupations frequently. This means that workers will have to respond to more labor market churn by learning new skills to enable them to transition from one job to another. There is also likely to be more pressure on governments to offer a social safety net for displaced workers unable to find new jobs quickly.

A 2018 OECD study of the risk of jobs being eliminated by automation in 32 countries found that a seventh of jobs were highly likely to be automated quickly and an additional third of current jobs could be eliminated with better machines and higher labor costs. Jobs at greatest risk of being eliminated include those in food preparation, construction, cleaning, and farming, sectors that employ many migrant workers in industrial countries.

AI can be weak, performing routine tasks, or strong, adapting over time

Source: Little Hoover Commission. 2018. Artificial Intelligence. Report 245.

Machine learning improves over time (spam filters) while deep learning adapts and learns (language translation)

Source: Little Hoover Commission. 2018. Artificial Intelligence. Report 245.

Over half of jobs in sectors that employ migrant workers could be eliminated by automation

Source: Nedelkoska Ljubica and Glenda Quintini. 2018. Automation, skills use and training. OECD.