Artificial intelligence (AI) is all around us, generating excitement about how it could increase prosperity and transform our lives in multiple ways. Yet the technology is also likely to be disruptive. Policymakers and businesses must therefore try to capture the full value of what AI has to offer, while avoiding the downside risks.
The idea of AI has been around for more than a half-century, and we have lived through previous periods of excitement followed by long stretches of disappointment - "AI winters" - when the tech didn't live up to the hype. But recent progress in AI algorithms and techniques, combined with a massive increase in computing power and an explosion in the amount of available data, has driven significant and tangible advances, promising to generate value for individuals, businesses, and society as a whole.
Companies are already applying AI techniques in sales and marketing to make personalised product recommendations to individual customers. And in manufacturing, AI is improving predictive maintenance by applying "deep learning" to high volumes of data from sensors. By deploying algorithms to detect anomalies, firms can reduce the downtime of machinery and equipment, from jet engines to assembly lines. Our research has highlighted hundreds of such business cases, which together have the potential to create between $3.5 trillion and $5.8 trillion in value per year.
AI also can contribute to economic growth by augmenting and substituting labour and capital inputs, spurring innovation, and boosting wealth creation and reinvestment. (AI will also create some negative externalities and transition costs, but these will be outweighed by its benefits.)
We estimate that AI and analytics could add as much as $13 trillion to total output by 2030, increasing the annual rate of global GDP (gross domestic product) growth by more than one percentage point. Furthermore, our research suggests that AI will have the greatest benefits if it focuses on innovation-led growth, and if its diffusion is accompanied by proactive management measures - in particular, retraining workers to give them the skills they will need to thrive in the new era.
As AI contributes to faster GDP growth, social welfare is also likely to increase. We estimate that AI and related technologies could improve welfare by 0.5-1% per year between now and 2030. That would be similar to the social impact of previous waves of technology adoption, including the information and communications technology revolution.
AI could help to improve many aspects of wellbeing, from job security and material living standards to education and environmental sustainability. Its biggest positive contribution to welfare may come in the areas of health and longevity: AI-driven drug discovery is several times faster than that based on conventional research. And AI-based traffic management can reduce the negative impact of air pollution on health by 3-15 per cent.
One of the most exciting aspects of AI is its potential to help address a wide range of social challenges. Although the technology is not a panacea, it could potentially help the world to meet all 17 of the United Nations Sustainable Development Goals (SDGs). AI applications that are currently being field-tested include efforts to assist with disaster-relief efforts, track smugglers (including human traffickers), and help blind people navigate their surroundings. And an AI disease-detection system can identify skin cancer as well as or even better than professional dermatologists can.
For all its potential, however, AI also poses substantial challenges that need to be addressed. The technologies themselves are still in the early stages of development, and more breakthroughs are needed to make them widely applicable. And there are considerable problems of data availability, which in turn affect the quality of AI models. One critical area of concern is the impact of AI and automation on work. Overall, we expect that there will be enough work for everyone, and that more jobs will be gained than lost as a result of the new technologies. But policymakers will need to manage significant transitions and challenges arising from AI adoption at national, regional, and local levels.
In the fastest automation-adoption scenario, up to 375 million workers worldwide will need to switch occupational categories by 2030, while some 75 million will be affected in a midpoint scenario. The nature of almost all jobs will change, as people interact more closely with smart machines in the workplace. That will require new skills, presenting companies and policymakers with the major challenge of training and retraining the workforce at scale. And as demand for high-skill jobs grows, low-skill workers could be left behind, resulting in increased wage and income inequality.
The diffusion of AI will also raise difficult ethical questions. Some of these will relate to the use and potential misuse of the technology in areas ranging from surveillance and military applications to social media and politics. Algorithms and the data used to train them may introduce new biases, or perpetuate and institutionalise existing types. Other critical concerns include data privacy and the use of personal information, cybersecurity, and "deep fakes" that could be used to manipulate election results or perpetrate large-scale fraud.
Despite these challenges, AI can generate tremendous value for us all, if policymakers and businesses act swiftly and smartly to capture its full benefits and mitigate the inevitable risks. The long-awaited "AI spring" may finally be arriving, but we will need to be prepared to manage its onset with care.
James Manyika is Chairman of the McKinsey Global Institute and a senior partner in McKinsey & Company's San Francisco office. Jacques Bughin is a director of the McKinsey Global Institute and a McKinsey senior partner based in Brussels.
Copyright: Project Syndicate, 2019.