Artificial Intelligence (AI) is often described as the fourth industrial revolution and compared in terms of its capacity to change the world with the steam engine and electricity – a comparison which the European Commission itself has endorsed in its communication “Artificial Intelligence for Europe”. Other than changing the world as we know it, AI is poised to create between $3.5T and $5.8T in value annually, according to a discussion paper put together by McKinsey.
Artificial Intelligence (AI) is often described as the fourth industrial revolution and compared in terms of its capacity to change the world with the steam engine and electricity – a comparison which the European Commission itself has endorsed in its communication “Artificial Intelligence for Europe”.[1] Other than changing the world as we know it, AI is poised to create between $3.5T and $5.8T in value annually, according to a discussion paper put together by McKinsey.[2]
It is therefore not surprising that, other than numerous corporations and start-ups, a number of states are starting to direct their attention to and invest their resources in AI. China’s declared aim is to be the leading power in AI by 2030[3], a title currently held – arguably – by the United States thanks to Silicon Valley; with Japan, South Korea and Canada also being increasingly active in this field.
This post is the first from a series of three which aims to present and analyse the concept of “European AI” and the efforts made by the European Union towards realising this goal. As the title suggest, it provides an overview of EU’s stance on AI by discussing the key takeaways from two papers put forward by the European Commission – the “AI for Europe” communication and the “Coordinated plan on AI”.
The other two posts will focus on the two deliverables released so far by the High-Level Expert Group on AI (AI-HLEG) – which the Commission has assembled - the first discussing the “Ethical Guidelines on the use of trustworthy AI” comparing them with ethical guidelines proposed in China and the US, while the other will analyse the main recommendations made in the second deliverable “Policy and investment recommendations for trustworthy AI”.
The whole series will focus on the efforts in the AI sector made by the EU as a whole, leaving aside the individual efforts of each member state. One of the reasons for doing so is that all EU member states, together with Norway, have already agreed to put forward a common strategy on AI by signing a “Declaration of Cooperation on AI” – which is discussed in some more detail below.
In his book, The rise of AI superpowers, Kai Fu Lee – a computer scientist and former president of Google China now turned venture capitalist investing in AI – expresses the view that the US and China are the only two AI “superpowers”. According to him, these two countries will be leading the AI revolution and ultimately profit from it most. Kai Fu Lee predicts that China will meet its goal of surpassing the US and become the leading AI superpower, while Europe will struggle to even win “the bronze medal”.[4]
That is not to say, however, that the EU is not alive to the impact that AI will have. The documents that will make the subjects of this series have all been released in the last 18 months, showing that the EU has made AI one of its priority. This has been confirmed once more by the political programme of the president elect of the Commission, Ursula von der Leyen, in which she promises that the new Commission will put forward a legislative proposal for a coordinated European approach on the human and ethical implications of AI in the first 100 days of its mandate. In the same programme, the president elect also promises that she will prioritise investments in AI both through the Multiannual Financial Framework and through the increased use of public-private partnerships.
EU’s intention to take the lead in “ethical AI” is clear and putting together a legislative proposal on this matter will certainly confirm its leadership in setting the agenda for the regulation of AI. Putting forward a legislative proposal on such a complex and technical topic is however a rather difficult task. It is therefore all but certain that in order to be successful in this pursuit the new Commission will have to draw inspiration from the work already done.
In this communication the Commission provides its definition of AI: “Artificial intelligence (AI) refers to systems that display intelligent behaviour by analysing their environment and taking actions – with some degree of autonomy – to achieve specific goals”, adding that AI can be purely software based or embedded in hardware.
Accurately defining AI can be a particularly difficult task, as a number of experts have argued.[5] Indeed, the AI HLEG put forward a somewhat different definition than the one above.[6] It would therefore not be wrong to argue that one of the most fascinating but also complex issues that the Commission will face in putting forward the promised legislative proposal will be formulating a suitable definition for AI that will successfully withstand the scrutiny of experts in the field.
The Commission then takes note of Europe’s position in the global race for AI development, noting that it lags behind in private investments which totalled around EUR 2.4-3.2billion in 2016, compared with the 6.5-9.7 billion in Asia and 12.1-18.6 billion in North America. Subsequently, and perhaps to some extent as a response to this situation, the communication then goes on to outline its desired strategy for AI, basing it on three pillars:
The Coordinated plan on AI builds upon the communication discussed above and the “Declaration of Cooperation on AI”, emphasising the need for all member states to work together in this field for maximum success and calling on them to ramp up their investments. Member states are also encouraged to develop their own national strategy on AI by mid-2019 by “building on the work done at the European level”. It is worth nothing that a number of member states have already missed this soft deadline, promising instead to prepare something by the end of the year.[7]
This plan also refines the concept of “European AI” and the path that will be taken to achieve it by firstly outlining two key features that European AI will have and secondly some strategies and objectives set in order to achieve it.
Two key features of “European AI” will be, according to this plan, “ethics by design” – in other words, ensuring a lack of bias and compliance with ethical and legal principles, including the GDPR and competition provisions – and “security by design”, whereby cybersecurity, the protection of victims and the facilitation of law enforcement activities will be taken into account from the beginning of the design process.
As mentioned above, these two deliverables will be analysed in depth in two future entries. Both of them require and deserve particular attention as they are likely to be very helpful in painting a picture of what “European AI” will look like and provide insights into the legislative proposal promised by the president elect.
This first post of the “Towards in European AI” has focused on setting out the context of this conversation and presenting the first steps taken towards the creation of “European AI” – setting out the strategy, presenting a coordinated plan once the members states agreed to work together, creating the AI HLEG and tasking it with drawing a clearer picture of what “European AI” should look like.
Read part II of the blog here.
[1] European Commission communication, “AI for Europe”, COM (2018) 237 final, of 25.04.2018.
[2] https://www.mckinsey.com/featured-insights/artificial-intelligence/notes-from-the-ai-frontier-applications-and-value-of-deep-learning
[3] State Council, State Council Notice on the Issuance of the New Generation Artificial Intelligence Development Plan (R. Creemers et al., Trans), 20 July 2017, available at https://www.newamerica.org/cybersecurity-initiative/digichina/blog/full-translation-chinas-new-generation-artificial-intelligence-development-plan-2017/.
[4] https://sifted.eu/articles/interview-google-kaifu-lee-ai-artificial-intelligence/
[5] See, for example, https://www.forbes.com/sites/bernardmarr/2018/02/14/the-key-definitions-of-artificial-intelligence-ai-that-explain-its-importance/#4a63ac164f5d.
[6] High-Level Expert Group on Artificial Intelligence, “A definition of AI: Main capabilities and scientific disciplines”, available here: https://ec.europa.eu/digital-single-market/en/news/definition-artificial-intelligence-main-capabilities-and-scientific-disciplines.
[7] See, for example, Ireland’s case - https://fora.ie/ireland-artificial-intelligence-europe-4750704-Aug2019/.