Countries competing to be leaders in the use of artificial intelligence face a common threat in the prospect that China could rise to dominate the field, which would put them at a disadvantage across key sectors of the economy. Given this challenge, and the risk that China will use unfair practices to achieve its goal, it may be time to strengthen the historic economic partnerships between North America and Europe with a new focus on AI. In particular, there may be opportunities to deepen research ties, coordinate government projects, promote global standards, and protect common values for their citizens and businesses. This kind of transatlantic investment and research on AI can help all partners make more rapid progress in embracing AI-driven automation and digitization.
On July 3, the Center for Data Innovation hosted a conversation with five experts to discuss the opportunities for collaboration on AI between Canada, the European Union, and the United States. As they each pursue AI strategies, even in the context of a “global AI race,” there may be opportunities within the transatlantic community to deepen research ties, coordinate government projects, promote global standards, and protect common values for their citizens and businesses.
When discussing the future directions for the transatlantic relation, it is important to reflect on where it presently stands at the political, research, and scientific levels. Eline Chivot, Senior Policy Analyst at the Center for Data Innovation, recalled that as stakeholders on both sides of the Atlantic will have to work together to shape how AI is developed and applied, this means that in particular, the United States and the EU will have to move past some of the difficulties their relationship is going through, as these may hinder deeper cooperation on digital policy.
According to Kristine Berzina, Senior Fellow at the Alliance for Securing Democracy and the German Marshall Fund of the United States, as much as the trade and security relationship between the United States and the EU is still very powerful and robust today, this relationship is not as healthy as it was four years ago. There is increasing distrust between both players, particularly on economic relations.
Despite this, common values and approaches on both sides of the Atlantic, including Canada, remain. These are democracies that share a fundamental belief in civil liberties and rights. The core underpinning of these values creates similar approaches to AI in the EU and the United States—despite rhetorical differences. Both the EU and the United States have are developing strategies for AI that look at ethical and socio-economic elements such as education, job security, and the prevention of discriminatory practices.
According to Wolfgang Wittke, Advisor to the Head of Secretariat of EUREKA (an intergovernmental research organization which includes 47 countries and the European Commission as a partner), existing transatlantic research ties are powerful in the sense that they remain stable over time. The transatlantic context and scientific cooperation do not change depending on political leadership. Scientific agencies vary in terms of budget and focus but over the last two years there has not been a major shift in their orientations.
Wittke asserted that the history and intensity of R&D cooperation between the United States, Canada, and the majority of EU member states are significant. There are plenty of opportunities to cooperate on AI, and an impressively high number of cooperation agreements which keep on increasing. The United States is by far the first scientific partner of the European Framework program, and of any EU member state. With Canada, there is a trilateral cooperation which is extremely rich as well. AI can easily and naturally be integrated into the transatlantic relationship as an area for cooperation, because of this legacy and this broad experience. However, for the EU to hold a strong transatlantic dialogue on AI, it must first identify where to set its focus on AI cooperation, beyond establishing regulatory bodies and considering standard-setting approaches.
There are various reasons why cooperation on AI in the transatlantic space is critical. Chivot mentioned the continuous, shared economic interest, for instance as the EU and the United States have the world’s largest and most integrated bilateral trade and investment relationship. Stéphane Lambert, Counselor, Head of Trade, Economic, Science and Technology Policy at the Mission of Canada to the EU added that the EU and Canada both believe in open science and strong international collaboration, which drives a deep and rich collaboration on science. The EU is Canada’s second largest market when it comes to R&D partnerships, and forty percent of the Canadian community of researchers’s international publications are with European researchers. There is a particularly strong incentive for Canada to partner with European consortiums and enhance bilateral cooperation to achieve a higher impact in research and position its research institutes.
Cate Nymann, Senior Manager, Government Affairs and Public Policy at Cisco, and Vice-Chair Digital Economy Committee at AmCham EU agreed that the transatlantic economy is much stronger when its stakeholders work together generally, and this certainly applies to AI. Businesses can play a critical role to facilitate cooperation on AI at this level.Chivot recalled that beyond economic ties, there is also a common interest between both sides of the Atlantic to address the prospect of a rising China which could soon dominate the AI field, putting the United States, Canada, and the EU at a disadvantage across key sectors of their economy.
Berzina noted that governments can use AI tools to either infringe upon liberties or ensure they integrate built-in, diverse norms. The latter should be the role of the transatlantic community and beyond, including all democracies, to make sure that the norms around AI will promote what their societies hold dear.
Expanding on the role of industry in the transatlantic industry, Nymann recalled that AI used to be perceived as a standalone technology indeed, while now companies are widely integrating AI into their applications, products, and services to enhance them. At the same time, companies acknowledge that they have a responsibility in using this technology, which remains “special” because of its powerful impact.
Lambert asserted that it is important to hold discussions to build confidence in products and trust in each other’s regulatory framework, ensuring that these frameworks are comparable and offer a similar level of protection. Regulation does not have to mean limitation, disruption, or restriction. Canada believes that further progress on AI can only be achieved with a strong, principle-based, multi-stakeholder governance framework over AI and a trusting relationship with businesses—much like the EU with its high-level expert group’s work. Canada and the EU can collaborate to promote it. For example, the EU supported Canada and France’s vision to establish an international panel on AI (IPAI) replicating the model of the International Panel on climate change (IPCC), which could act as a strong, global reference point on the ethical use of AI, ultimately leading to the adoption of global norms.
Berzina noted that the creation of additional bodies would distinguish AI as something in and of itself rather than something fully integrated. But AI is set to become like electricity, or the Internet—and there is no single intergovernmental body such as IPCC on the Internet. Creating any kind of body on top of existing organizations risks dissociating high-level discussions from the ways in which the technology is used in practice.
According to Nymann, the transatlantic private sector supports the conversations happening in this context, for instance as businesses develop their own ethics guidelines as well. It is however important to consider the diversity of businesses when developing these guidelines, the diversity of AI applications across sectors, and the distinction between B2B and B2C markets. Ethical guidelines and principle-based approaches will have to be translated to make them applicable in practice, through concrete standards for industry and technical recommendations tailored for companies’ engineers, researchers, and business developers.
Chivot mentioned that the EU is struggling to develop applications that will be relevant commercially and there is a disconnect between the amount of AI talent in the EU and AI funding. Nymann also noted the fragmentation of investment and spending on AI across the EU is an issue—the UK, France, and Germany are in the lead, far ahead of all other member states. If the EU loses focus on overcoming this issue, it risks finding itself in a situation where it has strong ethical guidelines but does not have the technology to apply them to.
In order to come from the point of “very good research to very good products,” Wittke stressed the importance to set the appropriate framework for the development and integration of this technology in the European society, beyond values, in the area of education and skills—for instance by integrating science, technology, engineering, and mathematics (STEM) into curriculums.
Lambert acknowledged that the shortage of AI skills is affecting industry. Canada is actively promoting its AI ecosystem on the international stage in this area. Highly skilled AI workers coming to Canada for a short-term project are exempt from having to obtain a working permit. For longer-term projects, the country ensures a service standard of two weeks for workers to obtain the visa and the working permit to accelerate the application process. This ensures that established companies in Canada grow in a predictable environment, with a stable access to global talent. The results are that the share of Canadian companies in the AI field is growing by 30 percent year over year, with now more than 800 companies and one the world’s fastest startup creation rate.
Chivot recalled that despite there being distinct benefits for nations that are hubs of AI development and adoption, the “AI race” is not a zero-sum game. To a certain extent, greater transatlantic cooperation could help push back some of China’s nationalist tendencies, and the EU, the United States, and Canada could also cooperate with China on AI.
According to Berzina, as research on these fields is strongly integrated, U.S. and European research institutions are already involved with Chinese companies. Global standards for applications, fostered by researchers in Europe, Canada, and the United States, can be those that have authority when working with Chinese companies.
Wittke agreed that both sides of the Atlantic can take a leadership on those aspects of legislation—ethical, social aspects, the implications for education, for our society and future. If transatlantic partners agree on these standards, China would be more likely to comply given the market access this could represent. But this requires a more collaborative framework that would also include Latin America and Africa, and requires Europe to speak with one voice on the global stage.
Summing up, the transatlantic relationship provides a strong and healthy basis for a collaboration in AI. All partners share common values and a mutual interest in sharing research capabilities and projects. Transatlantic investment and research on AI can help all partners make more rapid progress. They can be the driving force in setting standards in AI at the global level, through a multilateral framework, and provide a check on China’s quest for AI domination. The EU has strong assets to share with its partners—for instance, Canada particularly relies on European research partnerships—however it can learn from them how to overcome the lack of strong building blocks for AI, such as in AI talent retention. In addition, for this relationship to be efficient, the EU should settle on a clearer direction for AI and keep in mind that industry must be able to translate ethical considerations in practice.