Copyright: Brian Turner (Flickr)
Copyright: Brian Turner (Flickr)

After years of frustration with its relatively slow pace of progress, artificial intelligence (AI) is finally beginning to deliver on expectations. Depending on who you ask, this is either cause for optimism – or horror. Putting dystopian visions of robots taking over the world aside, most observers are worried about the implications of how a workforce composed increasingly of robots will transform society and the economy.

Bill Gates has now jumped into the fray, suggesting that governments should tax robots to fund training for people replaced by robots. Rather than having a knee-jerk reaction to news of advances in AI, however, we should take a wider view of how technological advances have impacted human welfare over the past two centuries. As long as we implement ways to help society adjust to the next revolution, we have more reason to welcome than fear the arrival of robots.

One reason for the increased focus on AI has been the significant strides that “deep learning” systems have made in recent years. AI has already been used with success in a range of fields, such as manufacturing, transportation, and logistics. Even more cognitive tasks can now be performed by computers. For instance, deep learning has been successfully employed to detect skin cancer better than dermatologists, to classify images faster than humans, and even to beat a trained professional at the ancient board game of Go.

Beyond image classification, data analysis, and pattern recognition, there is even reason to believe that artificial intelligence could one day be employed in the most human of professions, such as law. Last autumn, for instance, researchers announced they had created an algorithm that could predict the legitimacy of a complaint lodged at the European Court of Human Rights with 79% accuracy. The researchers said such technology could be used to streamline the judicial review process by quickly analyzing and prioritizing applications for the judges. These and other signs suggest that it is only a matter of time before robots could begin to supplant even high-level professionals.

Not unexpectedly, this prospect has set off alarm bells in the worlds of technology, economics, and politics. Physicist Stephen Hawking has warned that AI has the potential to destroy middle class jobs, widening the income gap and causing political instability. In 2013, two Oxford University researchers predicted that 54% of jobs in the EU were at risk of being eliminated due to increased computerization over the next two decades.

In the world of politics, most notably, French socialist presidential candidate Benoit Hamon has called for taxing robots and for using the new income to fund a 750€ monthly basic income for French citizens. Hamon has gone further than most policymakers, who have called such propositions too difficult to implement and costly. For example, Patrick Schwarzkopf, director of the Brussels-based EUnited Robotics trade association, said that a robot tax is a “solution to a problem that doesn’t exist,” citing the fact that high levels of employment and high robotization tend to be correlated. Backing up his statement, researchers from Deloitte found in 2015 that while technology contributed to the elimination of about 800,000 lower-skilled jobs between 2001 and 2015, it also helped to create nearly 3.5 million higher-skilled jobs.

In addition to the shaky evidence about the extent to which robots destroy jobs, there is wide disagreement about how to feasibly implement a robot tax. Such a tax would require a practical definition of robots, a way to assess how they create value, and a method for attributing profits to defined automation programs. Additionally, Hamon’s proposition to use the funds from a robot tax to pay for a universal basic income is problematic. Different municipalities, such as Kangasala, Finland, have been experimenting with universal basic income, but it remains to be seen whether such programs will have a net positive effect. While they offer ways to patch holes in the safety net and to encourage entrepreneurism among the unemployed, they are prohibitively expensive and run the risk of abetting idleness.

There’s no question that Gates, Hawking, and even Hamon have raised legitimate concerns about how society will grapple with the shifts wrought by automation and other technological advances. But rather than responding with untested solutions such as taxing robots or handing out cash, policymakers should invest more in proven methods of addressing economic change: education, retraining, and development, in tandem with a strong social safety net. For instance, teaching students how to relearn, rather than specializing in one topic, will help prepare them for the vast changes on the horizon. “Nanodegrees” that can be completed in a few months are another way to quickly retrain workers. And welfare programs, though perhaps not as generous as universal basic income, will assist those at risk of falling through the cracks.

In preparing for the robotic revolution, it also helps to revisit the failures and successes of the Industrial Revolution. The shift from an agricultural to an industrial economy was characterized by intense economic, societal, and political disruption, as old jobs were rendered obsolete, new roles were created, and workers’ wages only belated reflected increases in productivity. At the time, worried observers such as the economist David Ricardo were “convinced that the substitution of machines for human labour…may at the same time render the population redundant.”

But we can take hope from the fact that the human race has overcome these challenges and has emerged stronger for it: over the long term, the substitution of machines for humans has boosted productivity, created more highly-skilled jobs and new industries, and increased quality of living. Over the next century, we can look forward to more numerous and dramatic shifts, as AI could transform every aspect of our lives from the cars we drive to the way that scientists perform medical research.

The main difference this time around is that as the pace of technological change is far swifter than in the 19th century, economic shifts could arrive at a wrenching pace. Jobs will be rendered obsolete in a matter of months, not decades, and workers will have to learn new skills many times over in their lives. The answers to such seismic shifts, however, will likely not lie in schemes such as taxing robots or universal basic income, but in programs that help people adapt to jobs that are not destroyed, but redefined.