In late April, the European Commission announced legislation on Artificial Intelligence (AI). Although well past its self-imposed deadline, the AI act builds on the ideas of the earlier white paper, proposing a trustworthy Union framework for the technology. Making AI systems more accountable, as well as banning applications that pose an unacceptable risk, does help define the rules of the game in the European Union, as well as ensuring that personal liberties continue to be safeguarded. While there is no denying that a comprehensive regulation is the way forward, the EU’s approach to AI needs to double down on innovation and uptake of the technology, or risk losing out.
To succeed in the digital decade, a key requirement is to be at the forefront of the technological revolution. This means creating a favourable environment for innovation. Unfortunately, this is not yet the case for the EU. The numbers speak for themselves; out of the top 100 global start-ups, only five are based in the EU. By some estimates, private investment in AI in Europe was below €2bn in 2020.
“While there is no denying that a comprehensive regulation is the way forward, the EU’s approach to AI needs to double down on innovation and uptake of the technology, or risk losing out”
In China and the United States, the relevant figures are, respectively, five and ten times higher. Some initiatives included in the AI legislative proposal such as sandboxing as well as the classification of a number of AI applications as minimal or limited risk will, to a degree, help spur innovation. But that in itself is not enough. To help the sector, we need to look more broadly.
For EU companies, AI-related or otherwise, successfully scaling up requires access to finance. Unlike in other markets, EU start-ups, especially those in Eastern European Member States, do not have many alternatives to domestic bank loans. The COVID-19 pandemic has complicated the situation further.
While impossible to fix the situation overnight, the EU can step up its game by beefing up the VentureEU capital fund-of-funds and other initiatives. To promote the EU’s development of AI, education has to play an increasing part. The UK alone has three times more undergraduate AI study programmes than the 27 EU Member States combined. In some countries there are no such programmes at all. As interest in AI studies has increased, the offer has to be matched with demand.
Retaining AI scholars is another challenge. For that, research and development opportunities have to be strengthened therefore public funding, with or without private partnerships, is paramount. The Commission’s commitment of a combined €1bn per year for AI, from the Horizon Europe and Digital Europe programmes, is vastly insufficient at the EU level when compared to relevant spending in other major economies. On a national level, EU Member States have a unique opportunity to tap the Recovery and Resilience Facility for stimulating both education as well as R&D initiatives connected to AI.
Regarding the uptake of AI, a lot will rest on societal acceptance. The predominant focus on the dangerous applications of AI and undesirable outcomes alone will not foster trust. The updated Digital Education Action Plan, which rightly concentrates on improving digital skills, can play its part. Yet if we want to promote novel technologies, we must keep highlighting the benefits that AI can bring. To illustrate, the debate often focuses on the need to protect workforces from a dystopian future, where robots take over our jobs.
“The UK alone has three times more undergraduate AI study programmes than the 27 EU Member States combined. In some countries there are no such programmes at all. As interest in AI studies has increased, the offer has to be matched with demand”
In fact, the opposite is true. Without stimulating and embracing innovation, in 30 years’ time most workers will still be performing tasks that could have been done by robots. It was precisely the great strides in technological progress, the embracing of automation and the resulting productivity growth that has, in the past, vastly improved the quality and duration of life. An average professional would now need to work only a couple of months a year to replicate the average income levels from a century ago.
Lastly, a certain acceptance and tolerance of risk is a prerequisite to an AI-aided future. For example, driving an autonomous vehicle is unlikely to ever be zero risk, but existing advanced driving aid systems are vastly reducing the likelihood of an accident when compared to unaided driving. We have to admit that a risk-free society is a utopia, with or without AI. The world is changing, with AI one of the major technologies driving that change. While on the one hand, the role of AI should not be overestimated - artificial general intelligence is many decades away, if ever achievable - the EU cannot afford to become complacent. The AI regulation is the first step. The hard work is still ahead.