Last week, the European Commission proposed a European approach to AI. To gauge the utility of the stated approach, what better benchmarks to reference than the ‘proposed key actions’ outlined within the coordinated plan review itself?
- Align AI Policy to remove fragmentation and address global challenges.
- Act on AI strategies and programmes by implementing them fully and in a timely manner to ensure that the EU reaps the full benefits of first-mover adopter advantages
- Accelerate investments in AI technologies to drive resilient economic and social recovery facilitated by the uptake of new digital solutions.
Addressing Challenges in a Fragmented World
For actors within the international system who are seeing their power decline relative to both the lofty heights of their past and the continued dominance of the US and ascendant China, they are faced with three clear choices:
- Wed themselves to the United States or China; a mutually exclusive decision.
- Position themselves as non-aligned, attempting to maintain a foot in both camps pragmatically pushing areas of enlightened self-interest.
- Chart course by offering an alternative middle ground, free from the shackles of hegemonic competition.
It is too early for most advanced nations to implement option 1 for fear of backing the wrong horse and the past twelve months has shown the difficulties to which option 2 poses for the European Union. Despite the agreement in principle being agreed in December 2020 for the EU-China Comprehensive Agreement on Investment, as of May 3rd 2021 – the agreement has not been ratified through the European Parliament due to human rights concerns in Xinjiang. This week, the EU-U.S Future Forum will be held to ‘develop a new transatlantic agenda and momentum for EU-U.S cooperation. In the field of artificial intelligence, the European Union has sought to drive option 3 – however, as I shall allude, the shackles of hegemonic competition are never too far away.
Revisiting the alignment goals of 1) removing fragmentation and 2) addressing global challenges; one must question whether these aims are mutually reinforcing or mutually exclusive given the bloc’s relative stature.
Like several fragile global actors, Europe is now an inward-looking power. As in the financial sector, the differences within respective tech ecosystems in member states far outweigh their similarities. The EU needs to gain consensus and rapidly deploy a broad-church organisation chaired by the private sector with real-time industry knowledge and support from both the EU and national governments. The fact that national authorities will be responsible for assessing whether AI systems meet their obligations undermines both the approach and the potency of the language used.
The terminology utilised, explicitly referencing ‘EU Values’ as opposed to broader ‘liberal’ values is subtle but attempts to convey the fact that it is a bloc pushing for a uniquely identifiable variant of Western values, at arm’s length from the U.S and indeed the U.K. There is good reason to do so, to increase and promote harmonisation; but greater provision must be paid to the method of implementation if the approach is to have any chance of success.
A Global Pace Setter?
The fact that such a lengthy and considered document has been published by a preeminent global player is welcome. It is the first document of detail and scale by a true international actor. However, without adequately recognising the flaws in the assumptions of the paper, the follow-through has the potential to become hollow – two primary assumptions come to mind.
The first assumption relates to Europe’s relative strength in the AI global ecosystem – which was explicitly referenced numerous times within the approach. This misnomer can be dealt with rather swiftly, of the world’s top 20 most valuable tech firms, only 3 are domiciled in Europe. 12 are US domiciled with 4 domiciled in greater China – with the exclusion of state-owned companies and unquoted Huawei. The Commission ‘proposed that the Union invest in AI at least EUR1 billion’ through Digital Europe and Horizon Europe. This should be welcomed, right? To put this into perspective; China was estimated to have spent $17 billion in 2017, raising to circa $70 billion in 2020.
The second assumption relates to trust. Remarks by both Margrethe Vestager, Executive Vice President of the EC for ‘A Europe Fit for the Digital Age’ and Commission Chief Ursula Von der Leyen underline trust in data sharing and AI as paramount. In the February 2020 White Paper, Von Der Leyen underscored trustworthiness as a prerequisite for the uptake of digital technology. Trust is not a prerequisite for uptake if optionality to engage does not get put to the individual consuming the technology through voluntary means. The fact that so many will openly consume this blog via their smartphone without recognising their information is being collected through various datapoints is testament to such an assertion.
Herein lies the issue, there is a consensus around the current state of play; in so far as the contemporary governance framework for artificial intelligence and more broadly tech governance is broken, categorized by an anarchic self-regulatory framework dominated by largely American tech giants – but little appreciation of relative power structures. These announcements were made, without irony, against a backdrop whereby, one of the main beneficiaries of machine learning through data procurement, Apple, have put forward data optionality into the AI equation more succinctly than any government has managed to date through the developing IDFA debate.
By introducing an approach with such weak foundations against two basic assumptions, is to put forward an approach towards a horse that has not only bolted the gate but gladly returned to its stables, trampled over its trainer, and then proceeded to steal the sugar cubes.
Risk vs Reward
In the closing remarks of my last blog about Artificial Intelligence, I referenced the idea that only the creation of a defined set of norms, values and regulations that adequately balance the rights of the individual with the creation of an innovation system underpinned by clearly defined nomenclature would be the key for burgeoning actors in the sphere to upskill their position in the international market. The primary aims are twofold; to make the field accessible to upskill the working population and to set a framework of best practice in order increase investment in innovation.
Where Vestager noted in her speech that this was an approach that sought to outline the necessary steps to ‘boost investments and innovation’ – the clarity required to do business falls short in place of a clarion call of uncertainty.
For such a weighty document, attempts to bolster the prerequisite nomenclature were ambiguous beyond the basic introduction and the mechanisms through which unacceptable behaviour was policed noticeably woolly.
Furthermore, The White Paper came across as a document inherently punitive in nature; skewed towards the prevention of tech giant pillaging at the expense of grassroots, collaborative endeavours. As the U.S and the U.K are experiencing. efforts should be made to delineate requirements for ‘big tech’ and for innovative start-ups and scale-ups – the existence of COADEC in the United Kingdom with such influence on the Treasury is a prime example. It is one thing to protect citizens through the prohibition of unauthorised data mining, but another to stifle technological progress for the greater good.
The four-tier pyramid-like structure that sets out the EU’s AI risk appetite ranges from minimal risk to ‘unacceptable’. With the explicit exception of a definitive ban on real-time facial recognition in public places the framework is, on the face of it, deliberately ambiguous and intended to curb what the Commission sees as ‘unacceptable’ uses of AI. This creates uncertainty for tech companies already domiciled within the single market and wards off potential innovators from entering the Single Market for fear of arbitrary rulings under the auspices of risk.
In summary, and whilst it is important to note that this is only a proposed approach prior to EP and member state debate, it is an approach that overstates Europe’s existing position in the AI sphere and places too great a weight on the prevention of data infiltration by tech giants as opposed to the creation of innovative homegrown systems for the benefit of European citizens. In the very first instance, the European approach needs to double down on its focus to remove fragmentation within the single market and, for now, park its global ambitions.
Written by Harry Wilby, SAMI Associate
The views expressed are those of the author(s) and not necessarily of SAMI Consulting.
Achieve more by understanding what the future may bring. We bring skills developed over thirty years of international and national projects to create actionable, transformative strategy. Futures, foresight and scenario planning to make robust decisions in uncertain times. Find out more at www.samiconsulting.co.uk.
If you enjoyed this blog from SAMI Consulting, the home of scenario planning, please sign up for our monthly newsletter here and/or browse our website.
Image by Gordon Johnson from Pixabay
Very thought-provoking. Thanks
Interesting point about EU values as “a uniquely identifiable variant of Western values, at arm’s length from the U.S and indeed the U.K”. This is important as AI decisions will reflect the values they are built on. Chinese answers to the classic “trolley” problem could be different from Western ones.
Huw