This week I attended an event at the London School of Economics about Artificial Intelligence: What level has it reached? Is human-level AI a realistic possibility? And if it is achieved in the near future, what will the consequences be for humanity?
AI is progressing fast, driven by three mutually reinforcing technological advances: collection of “big data”, supercomputing, and machine learning algorithms. But while most discussion of AI (including at this event) tends to lead to speculation about whether or not machines will rule the world, or whether we will ever fall in love with robots, the more immediate concern is jobs.
In recent years, governments have scrambled to keep up with the rise of the “sharing economy” and firms like AirBnB and Uber have led to changes in tax and labour market regulation. It’s been a classic case of private sector innovation outpacing regulators. But the looming impact of AI makes the “sharing economy” pale in comparison. According to some estimates, half of American jobs today could be automated in next decade or two.
One has to wonder about the extent to which governments are prepared for an economic shock of this magnitude.
Of course, since the earliest stages of industrialization, technology has disrupted established industries and changed the ways we work – so in some sense this is nothing new. And governments welcome innovation because it drives productivity increases, and in the long run, increases in real wages.
But we shouldn’t forget that there are always winners and losers as part of this process, and sometimes the long run can be very long indeed – economic historians have shown that real wages didn’t rise in the UK during the industrial revolution for fifty years.
Managing the social and economic impacts of new technology is a crucial challenge for government. To do this well, they need to have an open and systematic conversation with the public about the likely impact of AI on jobs. The key questions the public are likely to have include: What types of jobs are likely to be automated, and what impact will this have on wage levels and inequality? How will the nature of work change? And what is the appropriate policy response, be it through the welfare system or labour market regulation?
Governments haven’t always covered themselves in glory in the ways in which they respond to new technology. It’s now well recognised that one of the key drivers of inequality in advanced economies over the last thirty years has been what economists call “skills biased technological change.” New technologies have led to more of us working in high and low skilled jobs, while there have been less jobs in the middle of the income distribution, driving increased wage inequality. With the benefit of hindsight, it’s easy to see many governments dropped the ball on this issue – had they better understood the labour market impacts of new technology, they might have done more to adapt policy settings to lean against this trend.
We should draw lessons from this for economic policy looking ahead. There are huge potential upsides to AI and automation, but if recent research is right, there are also significant risks. The sooner policymakers engage in a conversation about an appropriate policy response, the better.