As an AI lawmaker, Europe could recover its global lost ground

Giles Merritt concludes his two-part assessment of AI’s threats and promises for Europe with a look at the vexed problems of regulation.

The noisy debate over artificial intelligence (AI) is yielding remarkably little consensus as to whether Europe can catch up with the US and China, and what might be the consequences if it can’t.

Some say the ‘inventive step’ that defines intellectual property rights will be key, so that the owners of AI technologies will dictate terms to the users. Others argue that the deciding factor will instead be the extent to which AI is applied to complex operations in manufacturing and services.

It’s true that Europe lags behind in AI, both in its more humdrum form of super-computing and in the so-called Large Language Models (LLMs) that critics say may eventually outpace human creativity. Of the fifteen leading AI research outfits, thirteen are in the US. Meanwhile, China has barnstormed ahead with the widespread introduction of AI which, inter alia, keeps close tabs on its 1.4 billion citizens.

So far, the digital era has been disappointing in terms of wealth-creating productivity. Almost forty years ago MIT Nobel laureate Robert Solow famously remarked “I see the computer age everywhere except in the productivity statistics”. Nowadays called the Solow Paradox, it still holds true. In the decade 2010-20, productivity in Europe improved by a niggardly 0.53 a year. Services account for roughly four-fifths of economic activity, and productivity growth there is between 0.16 per cent and zero, depending on the sector.

We may nevertheless be on the threshold of rapid change. A new report by the research arm of the global consultancy McKinsey reckons that generative AI is on course to add 4 trillion dollars to the world economy – about the size of Germany’s GDP. As to non-generative AI along with automation in general, their combined economic boost may be as much as $11 trillion.

But not by tomorrow. The report’s authors say the 2030s will see its earliest impact, and warn that “AI is unlikely to trigger an economy-wide jump in productivity, or to support sustainable and inclusive growth if its use is left to market forces.”

In other words, it will be up to policymakers and their political masters to chart the course of AI’s introduction. As well as accelerating its development they must wrestle with regulatory policies that protect mankind from potentially adverse effects.

This is arguably where the European Union comes into its own. The EU has far more regulatory credibility around the world than either the US, which tends to be lax about reining in Big Business, or China whose rules are unlikely to find global acceptance. By the end of this year or early 2024, the EU’s Artificial Intelligence Act is to become law, after which there will be a two-year period for it to be adopted onto the statute books of the 27 member states.

Transparency is the leitmotif, with all AI actors and users in Europe to abide by an ethical Code of Conduct that’s backed up by penalty fines that could rise to €40 million or seven per cent of a company’s global turnover. Financial services and employment conditions are the Act’s backbone and risk management will be at its core.

The EU has been uncharacteristically speedy with its AI Act. The Commission proposed the outlines in April 2021 and it was adopted by the European Parliament this summer. In America, President Biden has just announced the US Safety Institute, which will work with a score of federal agencies to evaluate AI’s risks. For once, the US is trailing behind the EU.

China also has concerns over regulating AI. When high-level representatives from 28 countries met early this month at Bletchley Park in the UK, Beijing was represented by its Vice-Minister for Technology, Wu Zhaohui, who summed up the challenges very succinctly. He told the conference that AI is “uncertain, unexplainable and lacks transparency.”

Europe has earned a reputation for even-handed rules that exercise global influence, so it could be that its blueprint for AI will be copied around the world. But how does that help European businesses to recover from their slow start?

The answer is that this clear EU framework should encourage a much faster adoption of AI in both public and private sectors. AI has the potential to transform costly and inefficient bureaucracies as well as commercial operations. It could, for instance, streamline EU members’ healthcare and social security services, releasing hundreds of thousands of civil servants onto labour markets currently beset by manpower shortages.

In the private sector AI is expected to revolutionise professions like law, banking and accounting. As in manufacturing, an estimated 60-70 per cent of employees’ time is often taken up by tasks that AI can perform in minutes. It could be that the productivity improvements that have long eluded the EU would be met in a few short years. In that case, it’s to be hoped that a more resilient Europe will also have recovered much of its lost ground in the creation of AI products and processes.

The first article in this series – Europe’s AI weakness has a long history of missed opportunities – was published on 14 November. The views expressed in this Frankly Speaking op-ed reflect those of the author and not of Friends of Europe.

You may also like…

• #CriticalThinking: AI regulation: getting it right from the get-go, by Jamie Shea
• Friends of Europe: Towards a sustainable and digitalised EU economy
• #CriticalThinking: Thinking big for a better society, by Ben Wreschner