Following a tradition of tackling the complex questions affecting Europe’s security at the intersection of public policy, science and technology challenges, the European Leaders Network (ELN) and Salamanca Group brought together experts in the field of artificial intelligence to discuss the current AI landscape and the geostrategic and societal implications of its future implications for Europe.

Respecting the Chatham House Rule in order to encourage open debate, we looked at how Europe compares with the United States and China. This generated some key takeaways and an absolutely fascinating, wide-ranging and thought-provoking discussion.  We will dive more deeply into some of the issues in the New Year.

The conundrum of technological progress in a nutshell?

“Don’t be surprised when you end up where you don’t expect to be and the person who took you there  doesn’t quite know how it’s happened, either.”

Following a tradition of tackling the complex questions affecting Europe’s security at the intersection of public policy, science and technology challenges, the European Leaders Network (ELN) and Salamanca Group brought together experts in the field of artificial intelligence (AI), from large aerospace and defence businesses to global full-service banks, to discuss the current AI landscape and the geostrategic and societal implications of its future implications for Europe.

Respecting the Chatham House Rule in order to encourage open debate, we looked at how Europe compares with the United States and China. This generated some key takeaways and an absolutely fascinating, wide-ranging and thought-provoking discussion.  We will dive more deeply into some of the issues in the New Year.

Setting the scene

We considered artificial intelligence as it refers to current advances in machine learning through neural networks and deep learning. These advances matter because they are becoming exponential, with huge economic opportunities and with the potential to bring about a new industrial and social revolution, touching all aspects of human lives.

We acknowledged that the notion of a global AI “race” misleadingly implies that there will be an end point and a winner who crosses the metaphorical finish line first.

At the same time, there was consensus that there is global competition, that there are first adopter and market leader advantages, and that the United States of America, China and Europe are all striving to become the world leader in AI.

A lens through which to view the global AI race

Assuming the framework of a global competition for AI dominance, let’s look at the critical inputs to AI that help to determine leaders and laggards.

One of our panellists pointed to three key factors – talent, data, and computing hardware. Another underlined the roles that regulation and sovereignty play in pursuit of AI dominance:

• Like other new and emerging technological fields, developing AI crucially relies on talent from a tiny pool. Firms are being bought for the talent they contain, not their products.
• Data, as the foundation of all machine learning, is the most strategic asset in the global AI competition. Population and market size can make a huge difference.
• AI depends on computing power, so it needs access to hardware.
• Countries that want data, talent or technology for AI may encounter rising friction because other nations increasingly want to retain sovereign capability in this space.
• And regulation is both a hindrance that limits innovation and a safety net that mitigates threats and risks. Concerns over data collection and privacy, and fear of AI, have led some to conclude that, with such huge potential to impact human lives, AI cannot be left unregulated, even though regulation struggles to keep up with technological evolution.

A comparative look at the global AI race

But let’s look at how our experts judged each of the global contenders fare across these five factors before we attempt an assessment of Europe’s current position and conclude with a whole range of remaining questions and challenges:

Talent	At present, the US, thanks largely to the global reputation of Silicon Valley and continuing pull of the GAFA quartet (Google, Amazon, Facebook and Apple), alongside significant government and corporate investment in AI research and development, leads the world in recruiting and retaining AI talent.	China follows closely behind, benefitting hugely from its government declaring the development of AI a national priority. While China has lost talent to US universities, there are efforts to encourage those US-educated researchers and entrepreneurs back to their homeland.	Europe does reap the rewards of its strong education system and leading AI research but often struggles to encourage the talent in its healthy pool to move from academia into entrepreneurial and application-focused practical roles.
Data	Similarly to talent, the US leads the way as a result of both the smartphone revolution and more relaxed privacy laws offering access to diverse datasets.	While China does have the advantage of being similarly able to collect vast amounts of data across its population, it suffers from the homogeneity of its datasets which pose practical challenges in terms of training datasets for algorithms.	Europe, on the other hand, does have access to large and diverse datasets, but is constrained by market fragmentation across more than 28 separate countries with different cultures and languages, and strict data protection and privacy rules.
Computing Hardware	Dominating the microchip market, and having banned companies from exporting chips to China, the US comes out as the clear leader.	China, forced to reinvest in its own hardware development after being cut off from the US source it’s long relied on, is catching up slowly.	Europe lags behind in terms of its own hardware development, relying heavily on US-manufactured chips.
Regulation	While there are increasing calls in the US for tougher AI regulation, legislation currently exists only in draft form.	China is characterised by a near total absence of regulation affecting AI.	Europe, keen to emerge as the global leader in ethical AI, comes out as the regulatory frontrunner, having developed guidelines, frameworks and regulations.
Sovereignty	US export bans can be seen as a clear attempt to safeguard US AI advances from foreign competition.	China’s commitment to become an AI leader and investment in hardware capabilities signals preparedness to succeed in isolation, if required.	AI nationalism is on the rise within EU countries, driven by an overarching concern of losing ground to main players. An EU army is needed to pool sovereignty in order to punch above any individual members.

Europe in the global AI race

Europe is not dominating the AI landscape. It tends to focus on AI’s problems and threats, where China and the US have a simpler focus on opportunities.  It needs a more entrepreneurial academia. It doesn’t have a choice around sovereignty: either it fragments into AI nationalism or it scales up to succeed.

However, Europe is home to world-leading fundamental research.  It has talent, if it can retain it. European companies control niche AI assets.  The EU is, or can be, a global regulator. It has the potential to make AI trust, systems integrity, data privacy and ethical approaches its USPs.

While this doesn’t answer the question of whether ethical AI leadership without AI leadership per se is possible, it does offer Europe a way forward, provided the business model can be made to work.  London could become a global hub for data trading.  And Europe should invest in ‘small data’ – the increasingly efficient use of smaller data sets for machine learning.

Indeed, developing AI responsibly, grounded in ethical principles and rule of law, need not be a burden on research and investment but a stepping stone for Europe to drive this powerful technology forward. More than a technical decision, this is one of policy and vision which only Europe is able to realise and offer leadership on at this point in time.

What more do we need to look at?

So while Europe endeavours to position itself as an ethical AI leader, turning data privacy, trust and ethics into its very own USP, and perhaps even becoming a regulatory superpower, what other challenges do we all need to tackle as countries around the globe embrace AI technologies and a world without them becomes increasingly unthinkable?

1 | Trade. If you can have a set of machine-learning models that you know have been generated from good provenance, that has intrinsically a lot of value and becomes a tradeable asset. Should we change our focus to looking at which country might be first to create a legislative environment that facilitates the transfer of these assets to determine the real winner of the global AI race?

2 | System. What can we do practically to create a single system that encourages and makes possible meaningful collaboration between government, business and academia to realise new AI opportunities?

3 | Trust. How can we be sure that we can trust the outcomes (and decisions) that AI technologies will inevitably generate? How do we find ways of identifying outcomes that are generated because the inputs that drive them have been tampered with? How do we detect and respond to tainted AI models?

4 | Responsibility. Who is ultimately responsible for ethics in technology? Do we rely on what the public accepts (never mind that ‘the public’ will vary across cultures, and generations), on engineers’ desire to do the right thing, or on an external regulator that sets the framework on what should and should not be done? And what role should technology companies themselves play in driving this forward?

We hope to tackle these and more questions in depth at the next European Leaders Network and Salamanca Group expert discussion.

If you would like to attend our next ELN thinktank breakfast on AI, please contact: info@salamanca-group.com

Group Director
ShaRon Kedar

Europe in the global AI race