COVID-19 means EU must change its AI proposals

There is no shortage of talent brewing in AI in Europe and people with an entrepreneurial spirit to see them through. We have an ever-growing number of startups and the promise they show is second to none. However, we desperately need to create an environment for the companies and technologies of the future to develop and succeed in Europe – to provide jobs but also to help meet the challenges facing our planet. This has always been the case, but the gathering storm resulting from COVID-19 makes the need for the right environment more urgent than ever. 

Last year, when the proposals for new AI regulations in the EU were first published, dozens of startups across the “Digital 9” countries decided to form a coalition – the Digital Future for Europe – to respond in unison to the European Commission’s White Paper.

Digital Future for Europe is a group of startups, scale-ups, associations, think tanks and successful tech businesses from across the D9 nations. We have come together to embed a positive, ambitious and innovative digital agenda for the whole of Europe.

Our response to the Commission was clear: these proposals are positive and a good start, but nowhere near perfect. A heavy-handed approach to regulation could hold the sector back and diminish its chances to grow and become more competitive. Big tech companies may be discouraged from setting up camp and investing in Europe and the continent will become a barren ground for AI startups seeking to compete globally.

In order to avoid stagnating the sector, Europe must allow data to flow as freely as possible, help countries to address skills shortages, support SMEs to embrace the benefits of AI, and give the research and innovation community the tools, incentives and investment it needs to thrive.

It’s true – there are positives in the proposals: the Commission’s ambition to improve data access in Europe; the intention to promote the adoption of AI in the public sector; the acknowledgement of the importance of AI to the traditional economy; the will to create the right environment for increasing AI in the private sector; focus on skills; and the importance of not overburdening SMEs with regulation.

However, there are elements of the proposals that really worry us.

The definitions of AI and high-risk AI are too broad. The set of commitments for the development of certain kinds of AI systems, as they stand, would be too onerous for SMEs and startups. The human oversight proposals require careful thought so as not to defeat the point of many types of artificial intelligence, which are about reducing the need for human oversight of time-consuming and pattern-revealing tasks.

It’s also vital to recognise that while big tech companies may be able to adapt to these changes, startups and existing SMEs will find themselves in an impossible position: trying to figure out how to keep creating and innovating with their hands very much tied with a tonne of red tape behind their backs. We have submitted a full response to the Commission’s consultation, reflecting the concerns of coalition members, making it available on our website.

We are living through a global pandemic that has dangerously slowed down the economy, already creating a tougher environment for startups to thrive and SMEs and scaleups to stay afloat. In light of that, the organisation I lead, the Dutch Startup Association, recently launched a support program called ReStartup to help startups manage and recover from the COVID-19 crisis. 

We foresee a prolonged period of economic decline following the pandemic and expect effects for startups lasting well into 2021. The very fact that we are having to take these measures should be a warning sign to the European Commission that any initiatives that make the environment for these organisations even more fragile will amount to a risk definitely not worth taking.

The truth is, even though our startups and SMEs are brilliant and competitive, we are already lagging behind. As the COVID-19 crisis developed, for example, tech companies jumped to the fore and united efforts with other businesses, researchers and the wider AI-developing community to facilitate the sharing of data and speed up the analysis of the virus’ characteristics, spread, effects on the human body, and potential treatments and bases for vaccines. Most of those companies were the US and China-based – sadly, very few of these initiatives happened in Europe and they were all but dwarfed by projects launched in those countries.

The Dutch Startup Association’s goal is to turn the Netherlands into the startup State of Europe. And as members of the Digital Future for Europe coalition, we believe our own countries must do more to lead Europe in tech policy and we believe European policy must follow the lead of the most successful countries within the EU. This means doing more to support the success of the European tech ecosystem, recognising that we are entering an age where all sectors become digitised and affected by AI. 

There is no longer a “digital” part of the economy — we have an increasingly digitised and automated economy and companies which have traditionally driven Europe’s success must adapt to this reality.

Going forward, it will be crucial for the Commission to reconsider the definitions of AI and high-risk AI – as well as the commitments set out for the mitigation of harm – the potential impact of “before the event” regulations and the complications and unintended consequences of the human oversight requirements. These are some of the changes we would like to see:

First, we believe it would be more appropriate to have a definition that focuses on the way in which AI is a learning and adaptive technology, and the definition of “high-risk AI” should not only focus on the risk but should consider the probability of AI systems causing harm.

Second, the EC should not try to pre-emptively regulate AI systems and should instead focus on the problems of the present through the use of industry standards rather than statutory instruments. In addition, their insistence on a blanket requirement for human oversight defeats the point of many types of AI, which are about reducing the need for human oversight of many tasks.

Third, the focus on skills must include three things: educating the next generation for a digital era; supporting the current workforce to reskill, and ensuring Europe attracts and retains the talent its digital industries sorely need.

Finally, the EU must provide support for SMEs and startups, such as exemptions for startups and SMEs in their first year of operation to encourage innovation, plus provide incentives for existing businesses that want to use AI systems in their operations.

In the same way that Europe failed to ride the wave of the digital revolution that saw the rise of American-based tech giants such as Google, Facebook and Amazon, if it misses the current wave – the new tech revolution – its economy could be severely impaired in the years to come, as the exponential growth of AI in China and the US allows them to pull further away.

If the Commission can get the balance of regulation right, it will succeed in creating an environment that provides new jobs, attracts investment and allows Europe to compete with the US and China. Get this wrong and the burden of regulation – on top of the COVID-19 economic storm – could seriously damage Europe’s tech sector at a crucial time.

Guest post by Thomas Vles, Director of the Dutch Startup Association

Main image credits: Viacheslav Lopatin/Shutterstock