Remarks by Angel Gurría
OECD Secretary-General
9 September 2019 - London, UK
(As prepared for delivery)
Dear Dean, Professor Chandy, Distinguished Guests, Ladies and Gentlemen:
Thank you for your warm welcome. I would like to thank Professor Papaioannou and the Wheeler Institute for hosting today’s conversation on Artificial Intelligence and Sustainable Development – a topic at the heart of the OECD’s policy agenda.
Fifty years ago, the world watched in awe as the first humans landed on the moon. This remains one of the greatest feats of science, technology and innovation, and is a testament to the incredible power of human ingenuity, intelligence and courage.
Today, Artificial Intelligence – AI – is helping us to build on this achievement, by giving us the power to map the moon, to locate and count craters, and even to (virtually) moonwalk! In the not-too-distant future, AI is likely to propel us even further, to new frontiers in space.
Here on earth, AI has the potential to catalyse remarkable progress in our education, health, transport, social protection, communication and energy systems. It can also help us meet our global ambitions, including the Sustainable Development Goals (SDGs) and the Paris Agreement.
But AI is also fuelling anxieties and ethical concerns. Many fear that it will facilitate automated discrimination, by codifying existing biases from the analogue world into the digital world, including those related to gender, race or even the justice system. To realise the full potential of this promising technology, we need one critical ingredient – trust. And to build trust, we need human-centred AI that fosters sustainable development and inclusive human progress. Allow me to highlight a few priorities.
First, we must seize AI’s potential for productivity and scientific advances.
AI holds significant potential to drive productivity gains. It is helping people make better predictions and decisions, be they a doctor, a shop-floor manager or a farmer in the field. AI start ups attracted over 12% of all worldwide private equity investments in the first half of 2018, reflecting this economic potential.
AI can also increase the productivity of science. This is particularly relevant at a time when we face significant global challenges – a persistently sluggish global economy, growing trade tensions, climate change, and ageing societies, to name a few. More than a decade ago at a laboratory in Wales, a robot named Adam became the first machine to discover new scientific knowledge independently – a compound that works against a drug-resistant parasite that causes malaria.
But such advances are not automatic; significant complementary investments are needed to support them. Many firms do not even use relatively basic digital technologies, let alone AI. In fact, OECD analysis shows that only 11% of small firms perform big data analysis. Meanwhile, the growing prominence of AI in science is raising important policy questions around access to data and high-performance computing, intellectual property, and education.
Second, we must prepare for the transformation of the labour market and build skills.
For workers, technological change presents great opportunities. It can improve flexibility, productivity and earnings; and it can also reduce exposure to dangerous, unhealthy and tedious tasks.
But there are also risks. Across OECD countries, on average, around 14% of jobs are at a high risk of automation in the next 10 to 15 years, and a further 32% have a significant risk of disruption. AI will change, and perhaps accelerate, the profile of tasks that may be automated.
Exploratory work suggested that, in 2017, only 13% of OECD adults used, on a daily basis, literacy, numeracy and problem solving skills with higher proficiency than computers.
There are also risks of increasing inequalities. In a rapidly changing world, lifelong learning is an imperative for everyone, particularly low-skilled workers. They face a higher risk of job automation, and on average are 40 percentage points less likely to participate in adult learning than high-skilled workers in OECD countries.
To face such challenges, we must prepare and protect future generations. The OECD’s Learning Compass for 2030 outlines the need to adapt education so that our children become digitally aware citizens. This means going beyond literacy and numeracy, and fostering data and digital literacy. But having technical expertise and training a new generation of data scientists and digital experts is not enough. It will also be crucial to teach them social and emotional skills, as well as a sense of responsibility and ethics.
Third, we need to put AI at the service of the environment.
AI can accelerate the transition to a circular economy, reducing environmental pressures and the risks of raw material supply shocks. In fact, AI can help improve industrial production by making physical assets more intelligent and boosting data driven decision-making. This will enhance predictive maintenance, helping firms to extend product life, minimise waste, and optimise the performance of their systems and processes.
AI is also spurring innovative environmental policy approaches. For example, the UK Government is working with industry to develop an electronic waste tracking system that records all waste movements through the economy. The system focuses on boosting compliance, and reducing waste crime, by detecting illegal dumping and landfill avoidance.
However, the digitalisation of the circular economy can also have unintended environmental consequences. Rebound effects by increased levels of production and consumption of circular activities or lagging regulation are two examples of risks that require careful management.
The OECD is setting a course for human-centred AI.
The OECD recognises that actions at the margin will not be enough to harness AI for sustainable development. We need a comprehensive approach.
In May, we launched the OECD Recommendation on Artificial Intelligence to promote the responsible stewardship of trustworthy AI. This Recommendation – which represents the first-ever set of intergovernmental principles on AI – includes technical definitions, principles for responsible stewardship of trustworthy AI (including human-centred values and fairness, transparency and explainability, robustness and accountability), recommendations for national policies on trustworthy AI and advice for international co operation. Last July, in Osaka, G20 Leaders endorsed the G20 AI Principles, which are largely drawn from the OECD Recommendation on AI.
We will continue to advance multi-stakeholder collaboration on AI, with the launch of the OECD AI Policy Observatory later this year. This initiative will facilitate the development of evidence and practical guidance for policymakers and act as a hub for OECD work on AI, focusing on the interplay between AI and productivity, jobs, skills and trust.
The OECD is also working hard to address labour market challenges. The OECD’s 2019 Employment Outlook and our 2019 Skills Outlook put forward a Transition Agenda for a Future That Works For All. This calls for strengthening lifelong learning, social protection and collective bargaining – while targeting interventions on those who need it the most.
Ladies and Gentlemen:
Masayoshi Son, the founder of Soft Bank, said of AI, “If we misuse it, it’s a risk. If we use it in good spirits, it will be our partner for a better life.”
We must harness the efficiency and intelligence of AI to build a better world where growth is inherently sustainable, inclusive and beneficial to all. Let’s give people the tools to succeed in the world of tomorrow, let’s use this potential to protect our planet, and let’s harness progress and innovation, to make it people-centred.
The OECD looks forward to strengthening our engagement with the London Business School, as we work together to design, develop and deliver better AI policies for better lives. Thank you.
See also:
OECD work on Artificial Intelligence
