by Jasmine Kendall and Walter Pasquarelli

In this latest instalment of – IAIDL’ Artificial Intelligence strategy series, we turn our attention to the European Union’s White Paper On Artificial Intelligence – A European approach to excellence and trust. The EU’s AI strategy, coordinated by the Robotics and Artificial Intelligence unit, was completed this year, following a public consultation process. This blog considers some of the strategy’s strengths, and areas for further development.

When the EU finalised its AI strategy in February 2020, it expressed a strong commitment to the safe and responsible use of artificial intelligence. The bloc set out to prepare the region for the socio-economic consequences of digital transformation and to create an ethical framework for the use of new technologies. 

The EU uses its strategy to carefully consider the ethical debates which AI has sparked. This is to be applauded, but the document has important shortcomings. First, the EU needs to pay more attention to ensuring that the benefits of technological development will be distributed evenly between member states, as there is a risk that AI will exacerbate existing socio-economic divides within the region. Second, the EU seems to position itself as a rival to China and the USA. However, it has adopted a risk-averse approach to AI development, making it unlikely that it will successfully compete with these countries on their terms. If the EU is to strengthen its position as the regional champion of a safe and responsible AI, it is crucial that these doubts are addressed.

Strategy strengths: Safe and ethical, with a focus on cooperation

• Strong focus on ethics and responsibility

The EU’s AI strategy places significant emphasis on the ethical implications of the “fourth industrial revolution”, setting out to create a “unique ecosystem of trust” in which the EU will act as a regulatory body.

For some, the EU’s ‘safe and slow’ approach to the implementation of artificial intelligence is counterintuitive to the fast-paced nature of digital innovation. Google, for example, has argued that the EU’s insistence on respecting GDPR, its suggested risk assessment requirements for market entry, and its strict rules on dataset use, will “significantly hamper the development and availability of beneficial AI”, by limiting companies’ ability to innovate.

However, it is questionable whether fast-paced innovation is the biggest priority for the European Union’s AI strategy. Rather, the EU seems more committed to reducing the potential risks associated with the future of AI, with the strategy cautiously warning that AI could to lead “opaque decision-making, gender-based or other kinds of discrimination, intrusion in our private lives or [technologies] being used for criminal purposes”.

The EU’s AI strategy acknowledges the economic, social and environmental consequences which artificial intelligence could bring, reacting accordingly by suggesting a comprehensive regulatory framework. Whether this is restricting or reassuring will depend on a stakeholder’s own priorities.

• Fostering cooperation

Another particularly strong element of the EU’s AI strategy is its commitment towards fostering cooperation between the 27 Member States. In 2018, the EU published a Coordinated Action Plan on AI to accompany the broader strategy. Speaking about the action plan, the president of the EU’s Digital Single Market, Andrus Ansip, emphasised how the plan calls upon nations to coordinate resources and share open data.

By taking this cooperative approach, the EU’s AI strategy seeks to achieve maximum value for money, and avoid the unnecessary duplication of research and development in different countries. Cooperation also extends up to setting up a widespread common data infrastructure, GAIA-X, alongside regional partners in Germany and France, “a secure, federated system that meets the highest standards of digital sovereignty while promoting innovation”.

The collaborative nature of the strategy also extends to its creation process. Following the publication of an initial draft, the EU’s strategy was subject to a 4-month long public consultation period, which invited stakeholders from the public and private sectors and academia to comment and shape its content. Survey responses were then published openly. This model should be a point of reference for others looking to create open and synergetic AI strategies.

• Funding to orchestrate cooperation

Alongside the publication of its AI strategy, the EU announced plans to increase its annual investments in AI by 70%, reaching $1.5bn in total for the period 2018-2020. Whilst this positive shift is impressive, the EU recognises that it lags behind China and the USA when it comes to AI funding. Indeed, this aspect of the EU’s AI strategy has still been subject to criticism. Deutsche Bank Research argues that “significantly more financial firepower is needed” if the EU is to close the investment gap which separates Europe from China or the USA (playing into a narrative of international competition which we take issue with later in this piece).

However, it is vitally important to stress that large amounts of money invested in AI by national governments, or the European private sector, won’t be included in this total. As such, the EU sees this funding as “the glue linking the individual efforts” of its Member States, “with an expected impact much greater than the sum of its parts”. The EU’s strategy sets out to encourage European governments to invest more than EUR 20 billion per year in AI over the next decade, an impressive commitment which has the potential to be particularly meaningful for smaller Member States, but only if funding is distributed responsibly, an issue we address below.

Areas to develop: Ensuring even development, and avoiding misplaced competitiveness 

• Working towards the more even distribution of AI initiatives across Member States

Although the EU’s AI strategy places significant emphasis on implementing AI ethically and responsibly, the issue of unequal digital development across the Member States goes underexplored in the main White Paper and its accompanying reports.

The strategy does make repeated warnings about “the fragmentation of the single market”. hinting that Member States should not be left behind as others advance. Meanwhile, the EU’s Coordinated Plan on AI, published in 2018, has also encouraged all Member States to have an AI strategy by 2019, and proposes to monitor progress in this regard.

Yet these small acknowledgements do little to address the fact that AI has the very real potential to create a serious digital divide within Europe. EU Member States such as France and Germany are investing significant amounts of capital into digital innovation, focussing on attracting talent to emerging AI hubs such as Paris or Berlin, whilst other countries are still in the process of drafting a national AI strategy. 

Therefore, without careful governance from the EU, the benefits of AI are likely to be spread unevenly, both in terms of urban / rural gaps, and Europe’s existing geoeconomic divide, which separate the North Western Member States from some of their Southern counterparts.

The EU claims that its strategy will help it become a “global leader in innovation” and seeks to make Europe attractive to AI researchers and companies who might currently be drawn to Asia or North America. Yet this desire for Europe to become an international leader in AI has the potential to overshadow the region’s own inequalities, and as such, the EU’s commitment to implementing an ethical AI ecosystem.

• Misplaced competitiveness 

The EU’s AI strategy sets out an ambitious vision for the EU to become a world leader in artificial intelligence. Whilst this isn’t problematic per se, the strategy adopts a very competitive stance in relation to players outside of the union, such as the United States and China; a stance that is discordant with the kinds of progress that can be achieved through global cooperation. For instance, one section of the strategy states;

“ […] the race for global leadership is ongoing, and Europe offers significant potential, knowledge and expertise.”

Framing the global AI arena as a “race for global leadership” is misleading in two ways. Firstly, AI is not a zero sum race. The EU has more benefit from extending its collaborative stance to countries outside of the union, than from pitting itself against them. 

Secondly, countries in the EU operate in a completely different way to China, which makes direct comparison difficult. For instance, whilst China might be able to collect data more freely, privacy concerns in democratic societies are likely to restrict member states taking the same approach. As such, the EU’s strategy rightfully places significant emphasis on GDPR, and moving forward, should embrace this regional identity, as opposed to pitting itself against other presences in the global AI arena.

Overall takeaways:  More work needed from the champions of a responsible AI

The EU’s AI strategy places significant emphasis on creating a “human-centric” and collaborative artificial intelligence ecosystem, prioritising a risk-averse and heavily governed approach to digital transformation.

With this in mind, the narrative of international competition with Asia and North America which runs through the strategy is somewhat aimless. The EU’s priorities set out in the White Paper differ completely from those of China or the USA. As such, seeking to engage in direct competition with these players seems futile. Instead, the EU should seek to establish its identity in the global AI arena. Given the tone set out in its White Paper, we anticipate that the EU would frame itself as a champion of responsible and ethical AI, as opposed to advocating a fast-paced, experimental and risky course of action.

However, if the EU is to fully cement its status as a defender of responsible AI, then it also needs to properly acknowledge how new technologies could shape the future of Europe. This is crucial when it comes to assessing AI’s potential to exacerbate existing inequalities between Europe’s North and South, or rural and urban areas. As a unifying power, we believe that the EU has a responsibility to promote evenly spread development amongst its Member States.

In order to achieve this, we at – IAIDL call for the European Union to conduct more research into the uneven development that artificial intelligence is likely to bring to the region, which explores options for mitigating against these unwanted consequences of digital transformation.

After all, the EU’s White Paper seeks to “better the lives of all its citizens” by harnessing the power of AI and a data-driven economy. This will only be possible if technological innovation is expanded beyond the urban centres of Northern and Western Europe.

For further reading, take a look at – IAIDL’ blog “What makes a good AI strategy?”. Watch this space for more instalments of Oxford Insight’s AI strategy series