On 17 November 2016, the OECD held the 2016 Technology Foresight Forum on Artificial Intelligence (AI).
Background on artificial intelligence
OECD Technology Foresight Forums have been organised since 2005 to help identify opportunities and challenges for the Internet economy posed by technical developments. Forum topics have included cloud computing (2009), ICTs and green growth (2010) big data (2012) and The internet of things (2014).
Why focus on AI? Artificial intelligence (AI) and machine learning are rapidly permeating our economies and societies. AI already underpins over 50% of global financial transactions. Earlier this year, an AI programme won at the game of GO against one of the World's best players – a feat that experts thought would take at least ten more years to accomplish.
What is the Technology Foresight Forum? The purpose of the ‘technology foresight fora’ organised every 2 years by the Committee on Digital Policy (CDEP) is to help policy makers identify and understand opportunities and challenges for the digital economy posed by new technical developments. In 2016, the theme chosen by OECD Member countries, with strong support from Japan's Ministry of Internal Affairs and Communications (MIC), is Artificial Intelligence (AI).
What is AI? AI-pioneer Marvin Minsky defined AI as "the science of making machines do things that would require intelligence if done by men". The term AI is therefore used when machines perform human-like cognitive functions such as learning, understanding, reasoning and interacting. For example, machines understanding human speech, competing in strategic game systems, driving cars autonomously or interpreting complex data are currently considered to be AI applications.
What is new? Despite fluctuations in public awareness, AI has been making steady progress since its inception in the 1950s. Recent breakthroughs in both machine learning techniques, computational power, and systems design have resulted in a sudden high profile as the outputs of AI become more human-like. The technology that underlies much of the success of the world's highest market capital firms (Google, Apple, Microsoft) has been developed gradually, but the prevalence of data, computation, and communication has created a fundamental shift in the power and availability of artificial intelligence, and its impact on everyday lives. For example, Google learns how to translate content into different languages based on translated documents that are online and Facebook learns how to identify people on images based on its database of known users.
What value from AI? Significant investments in AI over recent years attest to the numerous opportunities that AI is opening up in many different fields and applications. A range of digital-world applications exists in areas such as social networking, advertising, customer service, finance, and healthcare. Numerous physical-world applications are also emerging in transportation (e.g. driverless cars and flying drones) and manufacturing. Since AI sees data and patterns at scales that humans cannot, it creates opportunities for cost reductions and efficiency gains. Estimates of the value that will be generated by AI are extremely high: as an illustration, Accenture estimates the labour productivity increases from AI at between 11% and 37% percent depending on the country by 2035.
What timeframe? There are two different ways of thinking about the current and future uses and impacts of artificial intelligence: 1) 'Applied AI' systems are specialised to accomplish specific problem-solving or reasoning tasks -- most applications in use today fit in this category and they account for the bulk of AI research; 2) In a more distant future, a hypothetical 'artificial general intelligence' (AGI) where machines might be capable of general intelligent action, like a human being. Such an AGI would raise a large set of issues, some of which have been illustrated in science fiction. The Forum will focus on 'Applied AI' systems.
What policy response to AI? A number of leading AI experts warn that the likely impact of AI in the years ahead is still underappreciated by policymakers and the public at large. It is widely claimed that artificial intelligence technology, combined with "big data" and with computing power, will transform entire sectors of the economy and lead to in-depth societal changes.
In their joint declaration in April 2016, G7 ICT Ministers gathering in Takamatsu, Kagawa agreed to facilitate ICT technology R&D including in the area of Artificial Intelligence and stressed the need to ensure that policy frameworks take into account the broader societal and economic implications of such technologies.
Will AI replace human labour? A notable concern is that AI will create challenges for employment by augmenting or replacing human labour, not only in manual work but also in highly-educated professions such as radiology (e.g. with advances in image and pattern recognition). Policy frameworks may require adjustments, for example to facilitate continuous education, training and skills development and to address risks of growing earning inequality and unemployment.
Who will control AI technology? Some experts also caution that AI technology may become controlled by a few powerful companies with access to large amounts of data and funding. They fear that small firms and universities will be increasingly unable to compete with large players. There are also concerns about geopolitical impacts, with AI likely to affect replace labour from developing countries disproportionately, while developed countries largely control the technology, patents and data.
How to minimise risks? Applications such as predictive policing and facial recognition may improve safety but may also raise risks to civil liberties if people are monitored and inferences made by machines are not transparent. In order to minimise such risks, policy frameworks may take into account fairness, transparency and accountability as well as safety and controllability.
Who is responsible for AI decisions? AI-driven automated decision-making also raises questions of responsibility and liability, for example in the case of life-threatening accidents that may include driverless cars.
The meeting was chaired by Mr. Jörgen Abild Andersen, Chair of the Committee on Digital Economy Policy.
1. Welcome remarks
Mr. Andrew Wyckoff, Director for Science, Technology and Innovation, OECD
2. Key Artificial Intelligence (AI) Developments and Applications: Today and Tomorrow
Following an introduction to AI, the panel discussed the state of AI development today including recent break-throughs and key applications and opportunities offered by AI in areas ranging from social networking and advertising to healthcare and transportation. Panellists will also be asked to share insight on key opportunities and challenges raised by AI and on public policy issues priorities in this area.
Mr. Yves Demazeau, Research Director, Centre national de la recherche scientifique (CNRS), France (Presentation)
Mr. Olivier Ezratty, Innovation Strategies Consulting (Presentation)
Mr. Jonathan Sage, Governmental Programmes Executive, EMEA lead on cyber security and cloud computing policy, IBM (Presentation)
Ms. Ophélie Gerullis, Public Policy Manager, Facebook (Presentation not available)
Ms. Lynette Webb, Senior Manager, European Policy Strategy, Google (Presentation)
Q&A and discussion - Forum speakers and Committee delegates
3. Artificial Intelligence and Society: the Challenges Ahead
Ms. Yuko Harayama, Executive Member, Council for Science and Technology Policy, Cabinet Office of Japan (via video link - Presentation)
4. Public Policy Considerations Raised by AI
The panel considered some of the public policy considerations raised by AI, notably its implications for jobs, skills and education, but also ethical concerns, liability and responsibility questions, geopolitical implications and social inclusion issues.
Mr. Susumu Hirano, Faculty of Policy Studies／Professor, Dean, Graduate School of Policy Studies, Chuo University (Presentation)
Ms. Joanna Bryson, Reader at University of Bath, and Affiliate, Center for Information Technology Policy at Princeton University (Presentation)
Mr. Cyrus Hodes, Director for Artificial Intelligence and Representative for the MENA Region, The Future Society @ Harvard Kennedy School of Government (Presentation)
Ms. Cornelia Kutterer, Director of Digital Policy, Microsoft EMEA (Presentation)
Mr. Tatsuya Kurosaka, Project Assistant Professor at Keio University Graduate School of Media and Governance (Presentation)
Q&A and discussion - Forum speakers and Committee delegates
5. Wrap-Up and Next Steps
Ms. Anne Carblanc, Head of OECD Division on Digital Economy Policy