Are superhuman computers set to steal our jobs, take over the world and perhaps even kill off humanity?
Such fears are commonplace in contemporary culture. Witness how dystopias about super-powerful rogue forms of artificial intelligence (AI), such as Person of Interest, Black Mirror and Ex Machina, have proliferated in recent years. But while scary science fiction may be entertaining, the reality of AI is rather more prosaic – and positive.
For good or ill, AI isn’t going to be running the world any time soon. But it could provide huge possibilities for progress, such as new life-saving medicines, time-freeing self-driving cars, and smart energy-saving homes.
By automating routine tasks such as checking contracts and monitoring stocks, it could take much of the drudgery out of office work, making many people much more productive. It could also help businesses and public administrations – notably the EU institutions – become more competent, cut costs and communicate better.
For sure, progress can also cause problems. Like previous new technologies, such as smartphones, AI will disrupt established ways of doing things, displace some jobs and pose ethical dilemmas.
But it would be a mistake to deny Europe the opportunities of AI by focusing solely on its potential downsides; far better to try to seize the benefits of innovation while managing its costs and risks.
The EU ought to embrace AI, while helping people to adjust and addressing ethical issues through appropriate regulation. Ursula von der Leyen, the new European Commission President, who took office on 1 December, has promised legislation on the latter within her first 100 days.
The benefits of Artificial Intelligence
AI is shorthand for digital technologies that enable computers to mimic human cognitive functions, such as “read” text and numbers, “see” and “recognise” people and objects, “hear” and “understand” sounds and speech and reply. It’s what powers virtual assistants such as Apple’s Siri, smart speakers such as Amazon’s Alexa and the facial-recognition software at automated passport gates.
Thanks to AI, computers can perform tasks that previously only human brains could do, such as “learn”, spot patterns, solve problems and make predictions.
As ever more data is collected, the computer power to process it multiplies and algorithms to interpret it improve. AI is increasingly able to learn from its mistakes and better its performance, enabling it to match and in some cases surpass human capabilities. In 1997, IBM’s Deep Blue supercomputer beat the reigning world champion at chess; twenty years later, Deep Mind’s AlphaGo bested the world number one at the Japanese game of Go, not by copying successful human strategies, but by devising its own better ones.
In many cases, though, AI helps to augment human skills: while it is now able to spot tumours that doctors miss, the most effective cancer diagnosis combines computer power and human expertise.
More broadly, AI can be part of the solution to Europe’s biggest economic challenge: its sluggish productivity growth. The value of what workers in the EU produce each hour increased by only 8.7% between 2010 and mid-2019.
That meagre long-term labour productivity growth of less than 1% a year determines the rate at which wages and living standards can rise, the scope for social spending and government investment in education, research and infrastructure, and ultimately the ability to sustain the societies and way of life that Europeans cherish.
Europe’s poor productivity performance also shapes the political context at both national and EU levels. When wages and social spending stagnate, discontent rises and distributional divides sharpen, both within and between EU countries.
Populist politicians may also successfully direct Europeans’ disenchantment at the EU. And when the EU economy underperforms, its weight in the world ebbs and with it the influence to shape global issues that Europeans care about, from climate change to trade and development. Thus the stakes could scarcely be higher.
Economic and political impact
Like previous technological advances, AI could boost Europe’s poor productivity performance – and even enable the EU to become a technological leader. Between 1990 and 2016, the diffusion of digital technologies contributed 0.4-0.6% a year to economic growth – around a third of the total – in nine European countries that consultants at McKinsey dub “digital front-runners”. Better still, AI has the potential to boost their growth by 1.2% a year until 2030.
The positive economic and political impact of AI could be transformational for Europe and its citizens. The challenge for EU policymakers is how to maximise its opportunities, while minimising and managing its costs and risks. The biggest concerns are about employment and ethics.
Even though employment is at record highs in Germany and some other EU countries, and unemployment continues to fall in most others, many people fear that AI will ultimately cause mass joblessness. One study reckoned that nearly half of American jobs were at high risk of automation by the mid-2030s.
But more comprehensive analyses suggest that far fewer jobs are at risk. The Organisation for Economic Cooperation and Development (OECD) estimates that only 9% of jobs in 21 OECD countries are automatable.
Whatever the correct figure, the automation of some jobs need not lead to job losses overall; new and better jobs will also be created, as happened with previous technological advances, such as the internet.
In practice, jobs tend to consist of a number of tasks, some of which may be readily automatable, and others not. Autonomous vehicles, for instance, may indeed do away with the need for many human drivers – although aeroplanes capable of flying on autopilot still have human pilots just in case.
In many cases, though, AI will complement human labour. For instance, AI may make it easier and faster for lawyers to sift through all the evidence in complex trials, making the justice system fairer and less costly.
At the same time, AI will help create new products and services, and hence new jobs for those who provide them. The explosive growth of social media has created new jobs for digital marketers and content moderators, for instance. Some now even earn a living playing computer games as a spectator sport.
The higher productivity – and thus the higher incomes – that AI generates will also increase demand for services that are not readily automatable, such as nursing, personal trainers, creative professionals and consultants. Personal care and business professionals are two of the categories of fastest employment growth.
In short, some jobs will disappear; others will change; and new ones will appear. Overall, the total number of jobs in the economy is unlikely to change. The bigger risk to employment would be not deploying AI widely; this could make many businesses in Europe globally uncompetitive.
Clearly, the transformation of the labour market and business organisation entails adjustment costs; replacing old ways of doing things with new and better ones always does. Often, these will be borne by businesses themselves. But governments also need to step in. They need to modernise education, upgrade skills and invest in life-long learning. They should also provide financial support for workers in transition, invest in supportive infrastructure such as 5G telecoms networks and regenerate depressed areas. While this is mostly a task for national governments, at EU level this requires a shift of priorities from subsidising inefficient sectors to investing in the future.
Thorny ethical issues also need to be addressed. If an accident is imminent, whose safety should a self-driving car prioritise? If an AI-powered system rejects an application for a bank loan, how do we ensure that it is not discriminating unfairly? How can we make positive use of facial-recognition software for verifying and safeguarding people’s identity, while preventing its abuse by governments or businesses? In addition to ethical considerations, the EU needs to consider how to manage other crucial issues such as security and privacy.
One way for EU policymakers to become better regulators would be to learn by doing. The EU institutions have huge potential to deploy AI to become more capable, competent, cost-effective and closer to citizens, as “Ever Cleverer Union”, a new report that I co-authored for Open Political Economy Network (OPEN), an international think tank, explains.
Making use of AI in their own work would also give EU policymakers and officials first-hand experience of its potential and pitfalls. That would help them strike the right balance between seizing the huge opportunities of AI for European citizens, businesses and public administrations and addressing any ethical, security and privacy concerns that are not properly addressed by existing regulations.
The EU faces a new world of geopolitical rivalry between the United States and China in which technology is a key battleground, and Europe is often a laggard. If Europeans want to shape their own destiny, they cannot afford to fall further behind. With appropriate government policies and sensible regulation, AI provides opportunities to make Europeans richer, freer, safer and better governed. The EU should seize them.
By Philippe Legrain