Fostering global collaboration towards trustworthy AI

On Tuesday 20 October, DIGITALEUROPE organised a closed roundtable to exchange views on artificial intelligence (AI) policy and governance. This online event brought together speakers and participants from across the world to share their perspectives on the next steps forward towards an innovation-friendly regulatory framework, including matters of AI definition and scope, the balance between strict legal requirements and soft-law or voluntary measures, and how to increase international and cross-border cooperation and alignment.

DIGITALEUROPE Director-General Cecilia Bonefeld-Dahl began the event by giving a short presentation on the status of Europe’s digital agenda and investment plan, where it stands on the global stage, and what approach to AI policy the EU will pursue. The floor was then opened for a discussion between representatives of the EU institutions, Member States, and international AI policy experts. The conversation was moderated by Alberto Di Felice, Director for Infrastructure, Privacy and Security.

Many points of agreement were found on shared values and principles. We all strive towards a trustworthy development and deployment of AI and emerging digital technologies. Trust and innovation can go hand in hand. When it comes to how to reach this goal, different regions have different approaches. The EU may be more ‘hands-on’ with specific regulatory requirements, while other countries try to build more on stakeholder dialogue and soft-law measures.

However, that should not stand in the way of seeking more global collaboration and alignment. Learning is not only something for machines, but for the public and the private sector too, as proved by the constructive discussions around beneficial AI policy and governance which were shared during the roundtable.

DIGITALEUROPE thanks all the participants to this AI Roundtable, and we are committed to continuing being a platform that brings together valuable expertise and perspectives from across the world.

Where the EU stands

It has been said before, jokingly but also with some level of truth, that when it comes to AI, the US has the money, China has the data, and Europe has the regulation. This prompts several considerations for the participants in the roundtable to consider.

• Firstly, how do we create an ambitious and innovation-friendly regulatory framework in the EU to ‘catch up’ and stimulate the development and deployment of AI?
• Secondly, how can the EU do that in a way that does not isolate it in the global sphere?
• What lessons can we learn from other governments and regions of the world, how can we more strongly collaborate with and inspire each other, and what values do we share?

It’s also in this context that the European Commission has allocated substantial funding for emerging technologies, such as AI, in the Multi-Annual Financial Framework. Accelerating the digital transformation should be a key element of the COVID-19 recovery plan and funding strategy.

The key is coherence

As outlined by the Commission’s AI White Paper, the primary goal of the EU is to develop coherent and future-proof legislation for AI, safeguarding European values while creating minimal burden to businesses. This takes the form of an ecosystem of excellence and an ecosystem of trust, two strands of a single AI strategy.

When it comes to regulation, the question now is examining where current legislation, such as the GDPR or Product Liability Directive, may not be sufficient or fit-for-purpose for AI and other new digital technologies. With big data come big responsibilities as well, and it’s important to ensure that European norms and values enshrined in the existing acquis are not overshadowed by new technologies.

Ideally, regulatory adaptations can be targeted towards ‘high-risk’ use cases for which the current regulatory and enforcement toolbox needs to be updated and expanded. This approach is supported by a large group of EU Member States, in a recent non-paper on AI.[1]

Soft-law[2], international cooperation and standards alignment are also of high importance to facilitate this strategy and improve AI uptake, a goal warmly supported by all participants and representatives at the Roundtable.

In the European Parliament, new reports have now been adopted looking at matters of AI and its relation to ethics, to safety and liability, as well as to Intellectual Property Rights and international law (including military use). An important goal here has been to stress the overall strategic relevance of AI, economically and politically. The use of Regulations, rather than Directives, is also encouraged to avoid regulatory divergence between the EU Member States or lead to a rift in the nascent AI (Digital) Single Market.

An additional key point is education and empowerment. Citizens and consumers need to be more aware of AI and algorithms, while companies and governments should strive towards transparent and trustworthy use of AI. Civil society plays an important mediator role between these actors.

After some delay, a new Special Committee for AI in the Digital Age has been established to increase coherence and to develop a unified position in the European Parliament ahead of legal proposals from the Commission.

AI has no borders

The international representatives at the Roundtable expressed a strong interest in ensuring global cooperation and alignment. AI and software development should not be prerogative of any particular country. Talent and skills are found everywhere, industrial supply chains are cross-border, and principles of fairness and sharing in the benefits of scientific progress are universal.

A common sentiment shared by the participants was the need to foster trust in AI, particularly to encourage companies and governments to use it. A lack of citizen trust will only be harmful. Organisations using AI need to earn their social license through accountable and trustworthy (business) practices.

Every country has recognised the potential of AI and is in the process of updating and upgrading their legal frameworks, investment plans and economic strategies.

For example, Australia has put forward a broad financial package to invest in emerging technologies, with a substantial slice for AI. However, this is only a first step, with a roadmap being developed to give further directions to research and investment.[3]

In Japan, multi-stakeholder consultations and dialogue have been on the agenda for the past years. Principle-based governance, developed in partnership between the public and the private sector, will improve adherence and be more agile and enforceable for emerging technologies.[4]

New-Zealand, for its part, is focusing on international cooperation and alignment. When it comes to AI policy and governance, it’s important to have a broad definition to not skip or overlook potential risks and benefits. Fairness and diversity, potentially harmed by historic biases in data-sets and software development, are seen as key points to be addressed.

Common principles and collaboration

What became evident during the Roundtable was the shared vision and importance put on principles of ethics (especially trust and accountability) as well as innovation, and the shared goal of harnessing the benefits of technological advancement.

How to reach these goals and how to measure them, on the other hand, show differences of approach. The EU is well-positioned in direct regulatory intervention and enforcement tools such as ex ante conformity assessments. Other regions prefer to work with principles and soft-law, or indirectly through, for example, public procurement requirements, sharing of best practices and guidelines. Of course, every country and region displays a mix of both approaches to different degrees.

Bringing together views and expertise from across the world only helps to reach those goals, as well as being a learning experience for all. International organisations such as the OECD play a key role here.[5] The new Global Partnership on AI aims to bridge the gap between theory and practice in AI, by bringing together multi-disciplinary expertise from the public and the private sector and by improving collaboration. This cross-pollination hopes to be a catalyst for beneficial, shared innovation.

[1] https://em.dk/media/13914/non-paper-innovative-and-trustworthy-ai-two-side-of-the-same-coin.pdf

[2] Such as the Trustworthy AI Assessment List, developed by the AI High-Level Expert Group: https://ec.europa.eu/digital-single-market/en/news/assessment-list-trustworthy-artificial-intelligence-altai-self-assessment

[3] See https://www.industry.gov.au/data-and-publications/building-australias-artificial-intelligence-capability and https://data61.csiro.au/en/Our-Research/Our-Work/AI-Roadmap

[4] https://www.meti.go.jp/press/2020/07/20200713001/20200713001-2.pdf

[5] See the OECD’s AI principles here: https://www.oecd.org/going-digital/ai/principles/